jizhizili / gfm Goto Github PK

View Code? Open in Web Editor NEW

901.0 43.0 137.0 39.66 MB

[IJCV 2022] Bridging Composite and Real: Towards End-to-end Deep Image Matting

License: MIT License

Python 91.42% Shell 8.58%

matting image-matting animal-matting composition segmentation datasets

gfm's People

Contributors

Stargazers

Watchers

Forkers

qiuweibin2005 rogalag xiamenwcy lvfuyu arrowyj zhigaloff undercontroller huanzs1123 uptodiff quan1995417 lepy hongkydeakin phanthaithuc d0nmoses nguyen-tri-nhan thorpham do-trong phongpirik nxt-27 nikhilnainta 8secz-johndpope amadeuzou yangyangsu29 99991 leedaga mlgdg limitmhw issamlaradji thuan-nguyenduc gengcauwong xrosliang braveryang shuangmei 1165048017 liuguoyou pkusnail xiaotie1005 baifree supuchun jqxx123 ericustc summonswar namtp-ai rogerzhangzz hzt314 atomeyang cv-ip shiyongde hzshonny mifk verusx fcgll520 timurguseynov danny1127 ebuty gongyong728125 anish9 marcojoao aureliuspatiens pengtianyu dontcryme muxuechen xinshao-wang prettygrass zongking123 charleshhy tenglike1997 dreamer121121 nick1211 tochange liuswot zhounature snoopybingo jingziyou c00renut yueyedeai zhaohanlin li0101 ml-cv-123 chenying99 eternalgogi92 olalaye guoz2013 wr19960001 luoyizhi516 fctfyi huyuejingling bruinxiong ci-uuu jamesz309 scott-mao h83dsf3 caihaunqai plcarles zhanghongyong123456 kduy zyzl1 kartikwar greatfeel peterzhousz

gfm's Issues

human matting

Hi, can this work be used on human matting?

Where can I find the data set you shared?

Thank you for sharing, your project is very interesting. I just read your paper carefully, and I want to use your data set to make some interesting projects (non-commercial). Where can I find the data set you shared?

Denoise dataset

Hello, Thank you for your excellent work. I met some difficulties when denoise. Can you provide your dataset after denoising? Thank you~

Train model on custom dataset

Hi!

Thanks so much for your work, the approach of segmenting images without trimaps is great 👍

Will you ever release the code that allows training on own custom data?

Thanks in advance!

Build own dataset

Hello，I want to ask, how does this build its own training set? Because I've tried to run this code with pictures, the effect is poor.

When I save the entire model, but I don't want to expose the network code, but others can deduce that

My tests had to contain code for the network structure, It's confusing to me.
Looking forward to your reply.

Environment setup instructions use Python 3.7.7

Conda python 3.7.7 is recommended but I do not see that package in Conda.

Failed to download the agreement document for BG-20K

Hi, I am trying to download the agreement document for BG-20k dataset. It seem to be timed out. Could you provide anthor way to download it. Thanks!

about the RSSN

hello, thanks for your excellent work.I'm interested in the performance of the composite image with RSSN.Do you have the code about the composite image?I want to test the effect.（sorry，I didn't find it in the project）
Looking forward to your reply

Boundary problem

I find that if i use the predict alpha multified with your initial image to get the color foreground, which is also the way used in your inference code, the final color foreground image will be surrounded by gray lines in boundary regions

Train speed is very slow!

I try to add ur focus branch to my own branch but i found that the train speed will be very slow, i found that in loss.backward() and optimizer.step() will cost much time about 50S for [32, 3, 512, 512] input.

Glance Decoder Loss

Hi @JizhiziLi , thanks a lot for this awesome work first of all!
I have a question about the GD Loss, please.
In the paper, you mention it is a Cross-Entropy Loss between the output of the GD and the ground truth (dependent on RoSTA).

According to the code you shared, glance_sigmoid is the output of GD and it has shape (3, H, W).
Given you implement the GFM-TT, I assume the ground truth here is the one-hot encoded trimap, which will have the same shape.
What confuses me is that you use a sigmoid activation as a very last step of GD.
Shouldn't you use SoftMax? The idea is to classify each pixel in one of 3 classes (foreground, background, unknown).
I am not sure I understand how sigmoid helps here.
Also, pytorch CELoss combines LogSoftmax and NLLLoss. Given the input has already gone through sigmoid, it cannot be passed to the CELoss, given LogSoftmax will be applied too.
Shall we "undo" the sigmoid and use CELoss?

Or probably I am overthinking this, and sigmoid is "just" an activation (as ReLU). So I can consider glance_sigmoid as the raw output of the decoder and safely pass it to CELoss.

Thanks again!

Will you publish datasets?

Comparison with other Neural Networks

Hey!

Thank you for sharing your work. It is always nice to have insights of STOTA ideas to handle certain computer vision tasks.
This is not really an issue but more like a request. I've noticed, while reading your paper, that you compare your result
to HAttMatting.

I, myself, have reimplemented their papers but my implementations contains 42M parameters, while you are claiming that it contains 109M params. I have used this piece of code

pytorch_total_params = sum(p.numel() for p in netG.parameters() if p.requires_grad)

only on the generator. To compute the number of trainable parameters.

Yet, My HAttMatting neural network is able to output quite precise alpha mask very quickly (due to pretrained model weights)

So my question is the following: can you share the pth/pt file (net the code) of your reimplementation of HAttMatting neural network because it looks a bit weird to me that it contains 109M parameters.

Thanks a lot

PM-10k - Dataset

Thank you for sharing your work with the community, it is very interesting. I've just finished reading your paper, and I think your ideas and the PM-10k dataset could help in a project I am working on. Where can I find download instructions for the PM-10k?

Test low-resolution wild images

I test the given pretrained model using the wild images from internet, but get bad results. masks are almost zero. Is something wrong?

any demo on human in-house matting result?

Error while concat decoder_4_f and encoder_3

I have loaded the model weight successfully. While doing inference, I'm facing an error in the Focus decoder while concatenating the result of decoder_4_f and encoder_3. I'm using r34 as a backend and TT as a rosta.

d4_f shape - [1, 256, 20, 28]
e3 shape - [1, 256, 20, 27]

Error:

File "gfm.py", line 411
d3_f = self.decoder3_f(torch.cat((d4_f, e3), 1))
RuntimeError: Sizes of tensors must match except in dimension 1. Expected size 28 but got size 27 for tensor number 1 in the list.

Versions:
OS - Ubuntu 20
python - 3.9.9
torch - 1.10.1
torchvision - 0.11.2

Training code

Will you release the training code?

can not find RIM code for train

Hello, I can find the code about GFM, but I can not find RIM, is that anything I miss? Can you help me?

colab demo not work

@JizhiziLi @chaimi2013
运行的是test_samples

how to add custom background image ?

i tested it on google collab, instead of white background I want to replace it with any image background.

can you please explain to me steps, or code part where I can do this changes @JizhiziLi

about the image normalize

hello, thanks for your excellent work. but i found that you did not do image normalize in your inference code. does that means you did not do the same thing in training code? the input image is the origin rgb image?

Looking forward to benchmark your work

Dear Jizhizili,
Thank you so much for your awesome work. I went through your demo images as well as your paper, very interesting results. Can't wait for the training code to really benchmark the capability of this architecture :p

Could this be adapted for human matting?

Hi! great work! As this looks like a great approach, i was wondering if it would also work for human matting in teh same way?

Thanks!

How to calculate the composition loss when gaussian noise is added to the composited image?

Should we add the same gaussian noise to the composited predicted image as the composited ground truth image?

PM-10K dataset

How to get high-quality annotated alpha mattes?

Hi, thank you for your great work!
I want to generate high-quality alpha mattes like AM-2k for other images, could you please tell me how to get them? Thanks!

about training code & human segmentation

When will the training code be released? Can this model be migrated to portrait segmentation?

There are black spots in the prediction results

subtract background

heya, thanks for sharing this, gave me alot of help trying to figure out segmentation. ive worked with watershed and contours earlier but a novice at best. wondering if you could push me in the right direction for enhancing the background matting as its picking up some noise (ref image). it clearly finds the fish and is really close to closing that mask,
i tried hough but no luck.

Using publicly available models did not achieve the performance stated in the paper.

I trained and tested on AM2k but did not achieve the performance stated in the paper. Even when using publicly available models on the AM2k test set, I still couldn't reach the performance reported in the paper. What could be the reason for this?

The number of blocks

hi,
Thank you for uploading the net code!
In the architecture diagram of the paper, there are 4 blocks(E0~E4) in the enconder (including deconder), but in the code, two additional stages( including 3 resnet blocks) are added, there are 6 stages in total, should the code prevail?

args.pred_choice means

What's the difference about pred_choice=1 pred_choice=2 pred_choice=3？
Why does pred_choice=3 require alpha and trimaps?
thanks ！

Unable to run in google colab

It looks like a file 'deploy_samples.sh' is missing for testing in google colab. Also the upload widget is not working in colab.

Expected date of Training code

Hi,
When is the paper review expected date?
Eagerly waiting for the training code to check out this exciting work

Why inference twice on the same image with different scales?

Will focus predictions from 1/2 images with glance predictions from 1/3 image produce better results?

Error with provided pretrained models.

Hello,

First of all, I would like to thank you for providing all of the data for your paper and congratulate you on the results. I followed the instructions for testing on my own sample images using the provided gfm_d121_tt.pth model. When loading, I get the following error:

Loading backbone: r34_2b
Loading rosta: TT
Loading model: models\gfm_d121_tt.pth
Predict choice: 3
Test strategy: HYBRID
Saving to the folder: 
Running on GPU with CUDA as True...
Traceback (most recent call last):
  File "core/test.py", line 312, in <module>
    load_model_and_deploy(args)
  File "core/test.py", line 295, in load_model_and_deploy
    model.load_state_dict(ckpt, strict=True)
  File "Anaconda3\envs\gfm\lib\site-packages\torch\nn\modules\module.py", line 1052, in load_state_dict
    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for GFM:
	Missing key(s) in state_dict: "resnet.conv1.weight", "resnet.bn1.weight", "resnet.bn1.bias", "resnet.bn1.running_mean", "resnet.bn1.running_var", "resnet.layer1.0.conv1.weight", "resnet.layer1.0.bn1.weight", "resnet.layer1.0.bn1.bias", "resnet.layer1.0.bn1.running_mean", "resnet.layer1.0.bn1.running_var", "resnet.layer1.0.conv2.weight", "resnet.layer1.0.bn2.weight", "resnet.layer1.0.bn2.bias", "resnet.layer1.0.bn2.running_mean", "resnet.layer1.0.bn2.running_var", "resnet.layer1.1.conv1.weight", "resnet.layer1.1.bn1.weight", "resnet.layer1.1.bn1.bias", "resnet.layer1.1.bn1.running_mean", "resnet.layer1.1.bn1.running_var", "resnet.layer1.1.conv2.weight", "resnet.layer1.1.bn2.weight", "resnet.layer1.1.bn2.bias", "resnet.layer1.1.bn2.running_mean", "resnet.layer1.1.bn2.running_var", "resnet.layer1.2.conv1.weight", "resnet.layer1.2.bn1.weight", "resnet.layer1.2.bn1.bias", "resnet.layer1.2.bn1.running_mean", "resnet.layer1.2.bn1.running_var", "resnet.layer1.2.conv2.weight", "resnet.layer1.2.bn2.weight", "resnet.layer1.2.bn2.bias", "resnet.layer1.2.bn2.running_mean", "resnet.layer1.2.bn2.running_var", "resnet.layer2.0.conv1.weight", "resnet.layer2.0.bn1.weight", "resnet.layer2.0.bn1.bias", "resnet.layer2.0.bn1.running_mean", "resnet.layer2.0.bn1.running_var", "resnet.layer2.0.conv2.weight", "resnet.layer2.0.bn2.weight", "resnet.layer2.0.bn2.bias", "resnet.layer2.0.bn2.running_mean", "resnet.layer2.0.bn2.running_var", "resnet.layer2.0.downsample.0.weight", "resnet.layer2.0.downsample.1.weight", "resnet.layer2.0.downsample.1.bias", "resnet.layer2.0.downsample.1.running_mean", "resnet.layer2.0.downsample.1.running_var", "resnet.layer2.1.conv1.weight", "resnet.layer2.1.bn1.weight", "resnet.layer2.1.bn1.bias", "resnet.layer2.1.bn1.running_mean", "resnet.layer2.1.bn1.running_var", "resnet.layer2.1.conv2.weight", "resnet.layer2.1.bn2.weight", "resnet.layer2.1.bn2.bias", "resnet.layer2.1.bn2.running_mean", "resnet.layer2.1.bn2.running_var", "resnet.layer2.2.conv1.weight", "resnet.layer2.2.bn1.weight", "resnet.layer2.2.bn1.bias", "resnet.layer2.2.bn1.running_mean", "resnet.layer2.2.bn1.running_var", "resnet.layer2.2.conv2.weight", "resnet.layer2.2.bn2.weight", "resnet.layer2.2.bn2.bias", "resnet.layer2.2.bn2.running_mean", "resnet.layer2.2.bn2.running_var", "resnet.layer2.3.conv1.weight", "resnet.layer2.3.bn1.weight", "resnet.layer2.3.bn1.bias", "resnet.layer2.3.bn1.running_mean", "resnet.layer2.3.bn1.running_var", "resnet.layer2.3.conv2.weight", "resnet.layer2.3.bn2.weight", "resnet.layer2.3.bn2.bias", "resnet.layer2.3.bn2.running_mean", "resnet.layer2.3.bn2.running_var", "resnet.layer3.0.conv1.weight", "resnet.layer3.0.bn1.weight", "resnet.layer3.0.bn1.bias", "resnet.layer3.0.bn1.running_mean", "resnet.layer3.0.bn1.running_var", "resnet.layer3.0.conv2.weight", "resnet.layer3.0.bn2.weight", "resnet.layer3.0.bn2.bias", "resnet.layer3.0.bn2.running_mean", "resnet.layer3.0.bn2.running_var", "resnet.layer3.0.downsample.0.weight", "resnet.layer3.0.downsample.1.weight", "resnet.layer3.0.downsample.1.bias", "resnet.layer3.0.downsample.1.running_mean", "resnet.layer3.0.downsample.1.running_var", "resnet.layer3.1.conv1.weight", "resnet.layer3.1.bn1.weight", "resnet.layer3.1.bn1.bias", "resnet.layer3.1.bn1.running_mean", "resnet.layer3.1.bn1.running_var", "resnet.layer3.1.conv2.weight", "resnet.layer3.1.bn2.weight", "resnet.layer3.1.bn2.bias", "resnet.layer3.1.bn2.running_mean", "resnet.layer3.1.bn2.running_var", "resnet.layer3.2.conv1.weight", "resnet.layer3.2.bn1.weight", "resnet.layer3.2.bn1.bias", "resnet.layer3.2.bn1.running_mean", "resnet.layer3.2.bn1.running_var", "resnet.layer3.2.conv2.weight", "resnet.layer3.2.bn2.weight", "resnet.layer3.2.bn2.bias", "resnet.layer3.2.bn2.running_mean", "resnet.layer3.2.bn2.running_var", "resnet.layer3.3.conv1.weight", "resnet.layer3.3.bn1.weight", "resnet.layer3.3.bn1.bias", "resnet.layer3.3.bn1.running_mean", "resnet.layer3.3.bn1.running_var", "resnet.layer3.3.conv2.weight", "resnet.layer3.3.bn2.weight", "resnet.layer3.3.bn2.bias", "resnet.layer3.3.bn2.running_mean", "resnet.layer3.3.bn2.running_var", "resnet.layer3.4.conv1.weight", "resnet.layer3.4.bn1.weight", "resnet.layer3.4.bn1.bias", "resnet.layer3.4.bn1.running_mean", "resnet.layer3.4.bn1.running_var", "resnet.layer3.4.conv2.weight", "resnet.layer3.4.bn2.weight", "resnet.layer3.4.bn2.bias", "resnet.layer3.4.bn2.running_mean", "resnet.layer3.4.bn2.running_var", "resnet.layer3.5.conv1.weight", "resnet.layer3.5.bn1.weight", "resnet.layer3.5.bn1.bias", "resnet.layer3.5.bn1.running_mean", "resnet.layer3.5.bn1.running_var", "resnet.layer3.5.conv2.weight", "resnet.layer3.5.bn2.weight", "resnet.layer3.5.bn2.bias", "resnet.layer3.5.bn2.running_mean", "resnet.layer3.5.bn2.running_var", "resnet.layer4.0.conv1.weight", "resnet.layer4.0.bn1.weight", "resnet.layer4.0.bn1.bias", "resnet.layer4.0.bn1.running_mean", "resnet.layer4.0.bn1.running_var", "resnet.layer4.0.conv2.weight", "resnet.layer4.0.bn2.weight", "resnet.layer4.0.bn2.bias", "resnet.layer4.0.bn2.running_mean", "resnet.layer4.0.bn2.running_var", "resnet.layer4.0.downsample.0.weight", "resnet.layer4.0.downsample.1.weight", "resnet.layer4.0.downsample.1.bias", "resnet.layer4.0.downsample.1.running_mean", "resnet.layer4.0.downsample.1.running_var", "resnet.layer4.1.conv1.weight", "resnet.layer4.1.bn1.weight", "resnet.layer4.1.bn1.bias", "resnet.layer4.1.bn1.running_mean", "resnet.layer4.1.bn1.running_var", "resnet.layer4.1.conv2.weight", "resnet.layer4.1.bn2.weight", "resnet.layer4.1.bn2.bias", "resnet.layer4.1.bn2.running_mean", "resnet.layer4.1.bn2.running_var", "resnet.layer4.2.conv1.weight", "resnet.layer4.2.bn1.weight", "resnet.layer4.2.bn1.bias", "resnet.layer4.2.bn1.running_mean", "resnet.layer4.2.bn1.running_var", "resnet.layer4.2.conv2.weight", "resnet.layer4.2.bn2.weight", "resnet.layer4.2.bn2.bias", "resnet.layer4.2.bn2.running_mean", "resnet.layer4.2.bn2.running_var", "resnet.fc.weight", "resnet.fc.bias", "encoder0.0.bias", "encoder1.0.conv1.weight", "encoder1.0.bn1.weight", "encoder1.0.bn1.bias", "encoder1.0.bn1.running_mean", "encoder1.0.bn1.running_var", "encoder1.0.conv2.weight", "encoder1.0.bn2.weight", "encoder1.0.bn2.bias", "encoder1.0.bn2.running_mean", "encoder1.0.bn2.running_var", "encoder1.1.conv1.weight", "encoder1.1.bn1.weight", "encoder1.1.bn1.bias", "encoder1.1.bn1.running_mean", "encoder1.1.bn1.running_var", "encoder1.1.conv2.weight", "encoder1.1.bn2.weight", "encoder1.1.bn2.bias", "encoder1.1.bn2.running_mean", "encoder1.1.bn2.running_var", "encoder1.2.conv1.weight", "encoder1.2.bn1.weight", "encoder1.2.bn1.bias", "encoder1.2.bn1.running_mean", "encoder1.2.bn1.running_var", "encoder1.2.conv2.weight", "encoder1.2.bn2.weight", "encoder1.2.bn2.bias", "encoder1.2.bn2.running_mean", "encoder1.2.bn2.running_var", "encoder2.0.conv1.weight", "encoder2.0.bn1.weight", "encoder2.0.bn1.bias", "encoder2.0.bn1.running_mean", "encoder2.0.bn1.running_var", "encoder2.0.conv2.weight", "encoder2.0.bn2.weight", "encoder2.0.bn2.bias", "encoder2.0.bn2.running_mean", "encoder2.0.bn2.running_var", "encoder2.0.downsample.0.weight", "encoder2.0.downsample.1.weight", "encoder2.0.downsample.1.bias", "encoder2.0.downsample.1.running_mean", "encoder2.0.downsample.1.running_var", "encoder2.1.conv1.weight", "encoder2.1.bn1.weight", "encoder2.1.bn1.bias", "encoder2.1.bn1.running_mean", "encoder2.1.bn1.running_var", "encoder2.1.conv2.weight", "encoder2.1.bn2.weight", "encoder2.1.bn2.bias", "encoder2.1.bn2.running_mean", "encoder2.1.bn2.running_var", "encoder2.2.conv1.weight", "encoder2.2.bn1.weight", "encoder2.2.bn1.bias", "encoder2.2.bn1.running_mean", "encoder2.2.bn1.running_var", "encoder2.2.conv2.weight", "encoder2.2.bn2.weight", "encoder2.2.bn2.bias", "encoder2.2.bn2.running_mean", "encoder2.2.bn2.running_var", "encoder2.3.conv1.weight", "encoder2.3.bn1.weight", "encoder2.3.bn1.bias", "encoder2.3.bn1.running_mean", "encoder2.3.bn1.running_var", "encoder2.3.conv2.weight", "encoder2.3.bn2.weight", "encoder2.3.bn2.bias", "encoder2.3.bn2.running_mean", "encoder2.3.bn2.running_var", "encoder3.0.conv1.weight", "encoder3.0.bn1.weight", "encoder3.0.bn1.bias", "encoder3.0.bn1.running_mean", "encoder3.0.bn1.running_var", "encoder3.0.conv2.weight", "encoder3.0.bn2.weight", "encoder3.0.bn2.bias", "encoder3.0.bn2.running_mean", "encoder3.0.bn2.running_var", "encoder3.0.downsample.0.weight", "encoder3.0.downsample.1.weight", "encoder3.0.downsample.1.bias", "encoder3.0.downsample.1.running_mean", "encoder3.0.downsample.1.running_var", "encoder3.1.conv1.weight", "encoder3.1.bn1.weight", "encoder3.1.bn1.bias", "encoder3.1.bn1.running_mean", "encoder3.1.bn1.running_var", "encoder3.1.conv2.weight", "encoder3.1.bn2.weight", "encoder3.1.bn2.bias", "encoder3.1.bn2.running_mean", "encoder3.1.bn2.running_var", "encoder3.2.conv1.weight", "encoder3.2.bn1.weight", "encoder3.2.bn1.bias", "encoder3.2.bn1.running_mean", "encoder3.2.bn1.running_var", "encoder3.2.conv2.weight", "encoder3.2.bn2.weight", "encoder3.2.bn2.bias", "encoder3.2.bn2.running_mean", "encoder3.2.bn2.running_var", "encoder3.3.conv1.weight", "encoder3.3.bn1.weight", "encoder3.3.bn1.bias", "encoder3.3.bn1.running_mean", "encoder3.3.bn1.running_var", "encoder3.3.conv2.weight", "encoder3.3.bn2.weight", "encoder3.3.bn2.bias", "encoder3.3.bn2.running_mean", "encoder3.3.bn2.running_var", "encoder3.4.conv1.weight", "encoder3.4.bn1.weight", "encoder3.4.bn1.bias", "encoder3.4.bn1.running_mean", "encoder3.4.bn1.running_var", "encoder3.4.conv2.weight", "encoder3.4.bn2.weight", "encoder3.4.bn2.bias", "encoder3.4.bn2.running_mean", "encoder3.4.bn2.running_var", "encoder3.5.conv1.weight", "encoder3.5.bn1.weight", "encoder3.5.bn1.bias", "encoder3.5.bn1.running_mean", "encoder3.5.bn1.running_var", "encoder3.5.conv2.weight", "encoder3.5.bn2.weight", "encoder3.5.bn2.bias", "encoder3.5.bn2.running_mean", "encoder3.5.bn2.running_var", "encoder4.0.conv1.weight", "encoder4.0.bn1.weight", "encoder4.0.bn1.bias", "encoder4.0.bn1.running_mean", "encoder4.0.bn1.running_var", "encoder4.0.conv2.weight", "encoder4.0.bn2.weight", "encoder4.0.bn2.bias", "encoder4.0.bn2.running_mean", "encoder4.0.bn2.running_var", "encoder4.0.downsample.0.weight", "encoder4.0.downsample.1.weight", "encoder4.0.downsample.1.bias", "encoder4.0.downsample.1.running_mean", "encoder4.0.downsample.1.running_var", "encoder4.1.conv1.weight", "encoder4.1.bn1.weight", "encoder4.1.bn1.bias", "encoder4.1.bn1.running_mean", "encoder4.1.bn1.running_var", "encoder4.1.conv2.weight", "encoder4.1.bn2.weight", "encoder4.1.bn2.bias", "encoder4.1.bn2.running_mean", "encoder4.1.bn2.running_var", "encoder4.2.conv1.weight", "encoder4.2.bn1.weight", "encoder4.2.bn1.bias", "encoder4.2.bn1.running_mean", "encoder4.2.bn1.running_var", "encoder4.2.conv2.weight", "encoder4.2.bn2.weight", "encoder4.2.bn2.bias", "encoder4.2.bn2.running_mean", "encoder4.2.bn2.running_var", "encoder5.1.conv1.weight", "encoder5.1.bn1.weight", "encoder5.1.bn1.bias", "encoder5.1.bn1.running_mean", "encoder5.1.bn1.running_var", "encoder5.1.conv2.weight", "encoder5.1.bn2.weight", "encoder5.1.bn2.bias", "encoder5.1.bn2.running_mean", "encoder5.1.bn2.running_var", "encoder5.2.conv1.weight", "encoder5.2.bn1.weight", "encoder5.2.bn1.bias", "encoder5.2.bn1.running_mean", "encoder5.2.bn1.running_var", "encoder5.2.conv2.weight", "encoder5.2.bn2.weight", "encoder5.2.bn2.bias", "encoder5.2.bn2.running_mean", "encoder5.2.bn2.running_var", "encoder5.3.conv1.weight", "encoder5.3.bn1.weight", "encoder5.3.bn1.bias", "encoder5.3.bn1.running_mean", "encoder5.3.bn1.running_var", "encoder5.3.conv2.weight", "encoder5.3.bn2.weight", "encoder5.3.bn2.bias", "encoder5.3.bn2.running_mean", "encoder5.3.bn2.running_var", "encoder6.1.conv1.weight", "encoder6.1.bn1.weight", "encoder6.1.bn1.bias", "encoder6.1.bn1.running_mean", "encoder6.1.bn1.running_var", "encoder6.1.conv2.weight", "encoder6.1.bn2.weight", "encoder6.1.bn2.bias", "encoder6.1.bn2.running_mean", "encoder6.1.bn2.running_var", "encoder6.2.conv1.weight", "encoder6.2.bn1.weight", "encoder6.2.bn1.bias", "encoder6.2.bn1.running_mean", "encoder6.2.bn1.running_var", "encoder6.2.conv2.weight", "encoder6.2.bn2.weight", "encoder6.2.bn2.bias", "encoder6.2.bn2.running_mean", "encoder6.2.bn2.running_var", "encoder6.3.conv1.weight", "encoder6.3.bn1.weight", "encoder6.3.bn1.bias", "encoder6.3.bn1.running_mean", "encoder6.3.bn1.running_var", "encoder6.3.conv2.weight", "encoder6.3.bn2.weight", "encoder6.3.bn2.bias", "encoder6.3.bn2.running_mean", "encoder6.3.bn2.running_var", "psp6.0.weight", "psp6.0.bias", "psp6.1.weight", "psp6.1.bias", "psp6.1.running_mean", "psp6.1.running_var", "psp5.0.weight", "psp5.0.bias", "psp5.1.weight", "psp5.1.bias", "psp5.1.running_mean", "psp5.1.running_var", "decoder6_g.0.weight", "decoder6_g.0.bias", "decoder6_g.1.weight", "decoder6_g.1.bias", "decoder6_g.1.running_mean", "decoder6_g.1.running_var", "decoder6_g.3.weight", "decoder6_g.3.bias", "decoder6_g.4.weight", "decoder6_g.4.bias", "decoder6_g.4.running_mean", "decoder6_g.4.running_var", "decoder6_g.6.weight", "decoder6_g.6.bias", "decoder6_g.7.weight", "decoder6_g.7.bias", "decoder6_g.7.running_mean", "decoder6_g.7.running_var", "decoder5_g.0.weight", "decoder5_g.0.bias", "decoder5_g.1.weight", "decoder5_g.1.bias", "decoder5_g.1.running_mean", "decoder5_g.1.running_var", "decoder5_g.3.weight", "decoder5_g.3.bias", "decoder5_g.4.weight", "decoder5_g.4.bias", "decoder5_g.4.running_mean", "decoder5_g.4.running_var", "decoder5_g.6.weight", "decoder5_g.6.bias", "decoder5_g.7.weight", "decoder5_g.7.bias", "decoder5_g.7.running_mean", "decoder5_g.7.running_var", "decoder6_f.0.weight", "decoder6_f.0.bias", "decoder6_f.1.weight", "decoder6_f.1.bias", "decoder6_f.1.running_mean", "decoder6_f.1.running_var", "decoder6_f.3.weight", "decoder6_f.3.bias", "decoder6_f.4.weight", "decoder6_f.4.bias", "decoder6_f.4.running_mean", "decoder6_f.4.running_var", "decoder6_f.6.weight", "decoder6_f.6.bias", "decoder6_f.7.weight", "decoder6_f.7.bias", "decoder6_f.7.running_mean", "decoder6_f.7.running_var", "decoder5_f.0.weight", "decoder5_f.0.bias", "decoder5_f.1.weight", "decoder5_f.1.bias", "decoder5_f.1.running_mean", "decoder5_f.1.running_var", "decoder5_f.3.weight", "decoder5_f.3.bias", "decoder5_f.4.weight", "decoder5_f.4.bias", "decoder5_f.4.running_mean", "decoder5_f.4.running_var", "decoder5_f.6.weight", "decoder5_f.6.bias", "decoder5_f.7.weight", "decoder5_f.7.bias", "decoder5_f.7.running_mean", "decoder5_f.7.running_var". 
	Unexpected key(s) in state_dict: "densenet.features.conv0.weight", "densenet.features.norm0.weight", "densenet.features.norm0.bias", "densenet.features.norm0.running_mean", "densenet.features.norm0.running_var", "densenet.features.norm0.num_batches_tracked", "densenet.features.denseblock1.denselayer1.norm1.weight", "densenet.features.denseblock1.denselayer1.norm1.bias", "densenet.features.denseblock1.denselayer1.norm1.running_mean", "densenet.features.denseblock1.denselayer1.norm1.running_var", "densenet.features.denseblock1.denselayer1.norm1.num_batches_tracked", "densenet.features.denseblock1.denselayer1.conv1.weight", "densenet.features.denseblock1.denselayer1.norm2.weight", "densenet.features.denseblock1.denselayer1.norm2.bias", "densenet.features.denseblock1.denselayer1.norm2.running_mean", "densenet.features.denseblock1.denselayer1.norm2.running_var", "densenet.features.denseblock1.denselayer1.norm2.num_batches_tracked", "densenet.features.denseblock1.denselayer1.conv2.weight", "densenet.features.denseblock1.denselayer2.norm1.weight", "densenet.features.denseblock1.denselayer2.norm1.bias", "densenet.features.denseblock1.denselayer2.norm1.running_mean", "densenet.features.denseblock1.denselayer2.norm1.running_var", "densenet.features.denseblock1.denselayer2.norm1.num_batches_tracked", "densenet.features.denseblock1.denselayer2.conv1.weight", "densenet.features.denseblock1.denselayer2.norm2.weight", "densenet.features.denseblock1.denselayer2.norm2.bias", "densenet.features.denseblock1.denselayer2.norm2.running_mean", "densenet.features.denseblock1.denselayer2.norm2.running_var", "densenet.features.denseblock1.denselayer2.norm2.num_batches_tracked", "densenet.features.denseblock1.denselayer2.conv2.weight", "densenet.features.denseblock1.denselayer3.norm1.weight", "densenet.features.denseblock1.denselayer3.norm1.bias", "densenet.features.denseblock1.denselayer3.norm1.running_mean", "densenet.features.denseblock1.denselayer3.norm1.running_var", "densenet.features.denseblock1.denselayer3.norm1.num_batches_tracked", "densenet.features.denseblock1.denselayer3.conv1.weight", "densenet.features.denseblock1.denselayer3.norm2.weight", "densenet.features.denseblock1.denselayer3.norm2.bias", "densenet.features.denseblock1.denselayer3.norm2.running_mean", "densenet.features.denseblock1.denselayer3.norm2.running_var", "densenet.features.denseblock1.denselayer3.norm2.num_batches_tracked", "densenet.features.denseblock1.denselayer3.conv2.weight", "densenet.features.denseblock1.denselayer4.norm1.weight", "densenet.features.denseblock1.denselayer4.norm1.bias", "densenet.features.denseblock1.denselayer4.norm1.running_mean", "densenet.features.denseblock1.denselayer4.norm1.running_var", "densenet.features.denseblock1.denselayer4.norm1.num_batches_tracked", "densenet.features.denseblock1.denselayer4.conv1.weight", "densenet.features.denseblock1.denselayer4.norm2.weight", "densenet.features.denseblock1.denselayer4.norm2.bias", "densenet.features.denseblock1.denselayer4.norm2.running_mean", "densenet.features.denseblock1.denselayer4.norm2.running_var", "densenet.features.denseblock1.denselayer4.norm2.num_batches_tracked", "densenet.features.denseblock1.denselayer4.conv2.weight", "densenet.features.denseblock1.denselayer5.norm1.weight", "densenet.features.denseblock1.denselayer5.norm1.bias", "densenet.features.denseblock1.denselayer5.norm1.running_mean", "densenet.features.denseblock1.denselayer5.norm1.running_var", "densenet.features.denseblock1.denselayer5.norm1.num_batches_tracked", "densenet.features.denseblock1.denselayer5.conv1.weight", "densenet.features.denseblock1.denselayer5.norm2.weight", "densenet.features.denseblock1.denselayer5.norm2.bias", "densenet.features.denseblock1.denselayer5.norm2.running_mean", "densenet.features.denseblock1.denselayer5.norm2.running_var", "densenet.features.denseblock1.denselayer5.norm2.num_batches_tracked", "densenet.features.denseblock1.denselayer5.conv2.weight", "densenet.features.denseblock1.denselayer6.norm1.weight", "densenet.features.denseblock1.denselayer6.norm1.bias", "densenet.features.denseblock1.denselayer6.norm1.running_mean", "densenet.features.denseblock1.denselayer6.norm1.running_var", "densenet.features.denseblock1.denselayer6.norm1.num_batches_tracked", "densenet.features.denseblock1.denselayer6.conv1.weight", "densenet.features.denseblock1.denselayer6.norm2.weight", "densenet.features.denseblock1.denselayer6.norm2.bias", "densenet.features.denseblock1.denselayer6.norm2.running_mean", "densenet.features.denseblock1.denselayer6.norm2.running_var", "densenet.features.denseblock1.denselayer6.norm2.num_batches_tracked", "densenet.features.denseblock1.denselayer6.conv2.weight", "densenet.features.transition1.norm.weight", "densenet.features.transition1.norm.bias", "densenet.features.transition1.norm.running_mean", "densenet.features.transition1.norm.running_var", "densenet.features.transition1.norm.num_batches_tracked", "densenet.features.transition1.conv.weight", "densenet.features.denseblock2.denselayer1.norm1.weight", "densenet.features.denseblock2.denselayer1.norm1.bias", "densenet.features.denseblock2.denselayer1.norm1.running_mean", "densenet.features.denseblock2.denselayer1.norm1.running_var", "densenet.features.denseblock2.denselayer1.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer1.conv1.weight", "densenet.features.denseblock2.denselayer1.norm2.weight", "densenet.features.denseblock2.denselayer1.norm2.bias", "densenet.features.denseblock2.denselayer1.norm2.running_mean", "densenet.features.denseblock2.denselayer1.norm2.running_var", "densenet.features.denseblock2.denselayer1.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer1.conv2.weight", "densenet.features.denseblock2.denselayer2.norm1.weight", "densenet.features.denseblock2.denselayer2.norm1.bias", "densenet.features.denseblock2.denselayer2.norm1.running_mean", "densenet.features.denseblock2.denselayer2.norm1.running_var", "densenet.features.denseblock2.denselayer2.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer2.conv1.weight", "densenet.features.denseblock2.denselayer2.norm2.weight", "densenet.features.denseblock2.denselayer2.norm2.bias", "densenet.features.denseblock2.denselayer2.norm2.running_mean", "densenet.features.denseblock2.denselayer2.norm2.running_var", "densenet.features.denseblock2.denselayer2.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer2.conv2.weight", "densenet.features.denseblock2.denselayer3.norm1.weight", "densenet.features.denseblock2.denselayer3.norm1.bias", "densenet.features.denseblock2.denselayer3.norm1.running_mean", "densenet.features.denseblock2.denselayer3.norm1.running_var", "densenet.features.denseblock2.denselayer3.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer3.conv1.weight", "densenet.features.denseblock2.denselayer3.norm2.weight", "densenet.features.denseblock2.denselayer3.norm2.bias", "densenet.features.denseblock2.denselayer3.norm2.running_mean", "densenet.features.denseblock2.denselayer3.norm2.running_var", "densenet.features.denseblock2.denselayer3.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer3.conv2.weight", "densenet.features.denseblock2.denselayer4.norm1.weight", "densenet.features.denseblock2.denselayer4.norm1.bias", "densenet.features.denseblock2.denselayer4.norm1.running_mean", "densenet.features.denseblock2.denselayer4.norm1.running_var", "densenet.features.denseblock2.denselayer4.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer4.conv1.weight", "densenet.features.denseblock2.denselayer4.norm2.weight", "densenet.features.denseblock2.denselayer4.norm2.bias", "densenet.features.denseblock2.denselayer4.norm2.running_mean", "densenet.features.denseblock2.denselayer4.norm2.running_var", "densenet.features.denseblock2.denselayer4.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer4.conv2.weight", "densenet.features.denseblock2.denselayer5.norm1.weight", "densenet.features.denseblock2.denselayer5.norm1.bias", "densenet.features.denseblock2.denselayer5.norm1.running_mean", "densenet.features.denseblock2.denselayer5.norm1.running_var", "densenet.features.denseblock2.denselayer5.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer5.conv1.weight", "densenet.features.denseblock2.denselayer5.norm2.weight", "densenet.features.denseblock2.denselayer5.norm2.bias", "densenet.features.denseblock2.denselayer5.norm2.running_mean", "densenet.features.denseblock2.denselayer5.norm2.running_var", "densenet.features.denseblock2.denselayer5.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer5.conv2.weight", "densenet.features.denseblock2.denselayer6.norm1.weight", "densenet.features.denseblock2.denselayer6.norm1.bias", "densenet.features.denseblock2.denselayer6.norm1.running_mean", "densenet.features.denseblock2.denselayer6.norm1.running_var", "densenet.features.denseblock2.denselayer6.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer6.conv1.weight", "densenet.features.denseblock2.denselayer6.norm2.weight", "densenet.features.denseblock2.denselayer6.norm2.bias", "densenet.features.denseblock2.denselayer6.norm2.running_mean", "densenet.features.denseblock2.denselayer6.norm2.running_var", "densenet.features.denseblock2.denselayer6.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer6.conv2.weight", "densenet.features.denseblock2.denselayer7.norm1.weight", "densenet.features.denseblock2.denselayer7.norm1.bias", "densenet.features.denseblock2.denselayer7.norm1.running_mean", "densenet.features.denseblock2.denselayer7.norm1.running_var", "densenet.features.denseblock2.denselayer7.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer7.conv1.weight", "densenet.features.denseblock2.denselayer7.norm2.weight", "densenet.features.denseblock2.denselayer7.norm2.bias", "densenet.features.denseblock2.denselayer7.norm2.running_mean", "densenet.features.denseblock2.denselayer7.norm2.running_var", "densenet.features.denseblock2.denselayer7.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer7.conv2.weight", "densenet.features.denseblock2.denselayer8.norm1.weight", "densenet.features.denseblock2.denselayer8.norm1.bias", "densenet.features.denseblock2.denselayer8.norm1.running_mean", "densenet.features.denseblock2.denselayer8.norm1.running_var", "densenet.features.denseblock2.denselayer8.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer8.conv1.weight", "densenet.features.denseblock2.denselayer8.norm2.weight", "densenet.features.denseblock2.denselayer8.norm2.bias", "densenet.features.denseblock2.denselayer8.norm2.running_mean", "densenet.features.denseblock2.denselayer8.norm2.running_var", "densenet.features.denseblock2.denselayer8.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer8.conv2.weight", "densenet.features.denseblock2.denselayer9.norm1.weight", "densenet.features.denseblock2.denselayer9.norm1.bias", "densenet.features.denseblock2.denselayer9.norm1.running_mean", "densenet.features.denseblock2.denselayer9.norm1.running_var", "densenet.features.denseblock2.denselayer9.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer9.conv1.weight", "densenet.features.denseblock2.denselayer9.norm2.weight", "densenet.features.denseblock2.denselayer9.norm2.bias", "densenet.features.denseblock2.denselayer9.norm2.running_mean", "densenet.features.denseblock2.denselayer9.norm2.running_var", "densenet.features.denseblock2.denselayer9.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer9.conv2.weight", "densenet.features.denseblock2.denselayer10.norm1.weight", "densenet.features.denseblock2.denselayer10.norm1.bias", "densenet.features.denseblock2.denselayer10.norm1.running_mean", "densenet.features.denseblock2.denselayer10.norm1.running_var", "densenet.features.denseblock2.denselayer10.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer10.conv1.weight", "densenet.features.denseblock2.denselayer10.norm2.weight", "densenet.features.denseblock2.denselayer10.norm2.bias", "densenet.features.denseblock2.denselayer10.norm2.running_mean", "densenet.features.denseblock2.denselayer10.norm2.running_var", "densenet.features.denseblock2.denselayer10.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer10.conv2.weight", "densenet.features.denseblock2.denselayer11.norm1.weight", "densenet.features.denseblock2.denselayer11.norm1.bias", "densenet.features.denseblock2.denselayer11.norm1.running_mean", "densenet.features.denseblock2.denselayer11.norm1.running_var", "densenet.features.denseblock2.denselayer11.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer11.conv1.weight", "densenet.features.denseblock2.denselayer11.norm2.weight", "densenet.features.denseblock2.denselayer11.norm2.bias", "densenet.features.denseblock2.denselayer11.norm2.running_mean", "densenet.features.denseblock2.denselayer11.norm2.running_var", "densenet.features.denseblock2.denselayer11.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer11.conv2.weight", "densenet.features.denseblock2.denselayer12.norm1.weight", "densenet.features.denseblock2.denselayer12.norm1.bias", "densenet.features.denseblock2.denselayer12.norm1.running_mean", "densenet.features.denseblock2.denselayer12.norm1.running_var", "densenet.features.denseblock2.denselayer12.norm1.num_batches_tracked", "densenet.features.denseblock2.denselayer12.conv1.weight", "densenet.features.denseblock2.denselayer12.norm2.weight", "densenet.features.denseblock2.denselayer12.norm2.bias", "densenet.features.denseblock2.denselayer12.norm2.running_mean", "densenet.features.denseblock2.denselayer12.norm2.running_var", "densenet.features.denseblock2.denselayer12.norm2.num_batches_tracked", "densenet.features.denseblock2.denselayer12.conv2.weight", "densenet.features.transition2.norm.weight", "densenet.features.transition2.norm.bias", "densenet.features.transition2.norm.running_mean", "densenet.features.transition2.norm.running_var", "densenet.features.transition2.norm.num_batches_tracked", "densenet.features.transition2.conv.weight", "densenet.features.denseblock3.denselayer1.norm1.weight", "densenet.features.denseblock3.denselayer1.norm1.bias", "densenet.features.denseblock3.denselayer1.norm1.running_mean", "densenet.features.denseblock3.denselayer1.norm1.running_var", "densenet.features.denseblock3.denselayer1.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer1.conv1.weight", "densenet.features.denseblock3.denselayer1.norm2.weight", "densenet.features.denseblock3.denselayer1.norm2.bias", "densenet.features.denseblock3.denselayer1.norm2.running_mean", "densenet.features.denseblock3.denselayer1.norm2.running_var", "densenet.features.denseblock3.denselayer1.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer1.conv2.weight", "densenet.features.denseblock3.denselayer2.norm1.weight", "densenet.features.denseblock3.denselayer2.norm1.bias", "densenet.features.denseblock3.denselayer2.norm1.running_mean", "densenet.features.denseblock3.denselayer2.norm1.running_var", "densenet.features.denseblock3.denselayer2.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer2.conv1.weight", "densenet.features.denseblock3.denselayer2.norm2.weight", "densenet.features.denseblock3.denselayer2.norm2.bias", "densenet.features.denseblock3.denselayer2.norm2.running_mean", "densenet.features.denseblock3.denselayer2.norm2.running_var", "densenet.features.denseblock3.denselayer2.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer2.conv2.weight", "densenet.features.denseblock3.denselayer3.norm1.weight", "densenet.features.denseblock3.denselayer3.norm1.bias", "densenet.features.denseblock3.denselayer3.norm1.running_mean", "densenet.features.denseblock3.denselayer3.norm1.running_var", "densenet.features.denseblock3.denselayer3.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer3.conv1.weight", "densenet.features.denseblock3.denselayer3.norm2.weight", "densenet.features.denseblock3.denselayer3.norm2.bias", "densenet.features.denseblock3.denselayer3.norm2.running_mean", "densenet.features.denseblock3.denselayer3.norm2.running_var", "densenet.features.denseblock3.denselayer3.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer3.conv2.weight", "densenet.features.denseblock3.denselayer4.norm1.weight", "densenet.features.denseblock3.denselayer4.norm1.bias", "densenet.features.denseblock3.denselayer4.norm1.running_mean", "densenet.features.denseblock3.denselayer4.norm1.running_var", "densenet.features.denseblock3.denselayer4.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer4.conv1.weight", "densenet.features.denseblock3.denselayer4.norm2.weight", "densenet.features.denseblock3.denselayer4.norm2.bias", "densenet.features.denseblock3.denselayer4.norm2.running_mean", "densenet.features.denseblock3.denselayer4.norm2.running_var", "densenet.features.denseblock3.denselayer4.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer4.conv2.weight", "densenet.features.denseblock3.denselayer5.norm1.weight", "densenet.features.denseblock3.denselayer5.norm1.bias", "densenet.features.denseblock3.denselayer5.norm1.running_mean", "densenet.features.denseblock3.denselayer5.norm1.running_var", "densenet.features.denseblock3.denselayer5.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer5.conv1.weight", "densenet.features.denseblock3.denselayer5.norm2.weight", "densenet.features.denseblock3.denselayer5.norm2.bias", "densenet.features.denseblock3.denselayer5.norm2.running_mean", "densenet.features.denseblock3.denselayer5.norm2.running_var", "densenet.features.denseblock3.denselayer5.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer5.conv2.weight", "densenet.features.denseblock3.denselayer6.norm1.weight", "densenet.features.denseblock3.denselayer6.norm1.bias", "densenet.features.denseblock3.denselayer6.norm1.running_mean", "densenet.features.denseblock3.denselayer6.norm1.running_var", "densenet.features.denseblock3.denselayer6.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer6.conv1.weight", "densenet.features.denseblock3.denselayer6.norm2.weight", "densenet.features.denseblock3.denselayer6.norm2.bias", "densenet.features.denseblock3.denselayer6.norm2.running_mean", "densenet.features.denseblock3.denselayer6.norm2.running_var", "densenet.features.denseblock3.denselayer6.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer6.conv2.weight", "densenet.features.denseblock3.denselayer7.norm1.weight", "densenet.features.denseblock3.denselayer7.norm1.bias", "densenet.features.denseblock3.denselayer7.norm1.running_mean", "densenet.features.denseblock3.denselayer7.norm1.running_var", "densenet.features.denseblock3.denselayer7.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer7.conv1.weight", "densenet.features.denseblock3.denselayer7.norm2.weight", "densenet.features.denseblock3.denselayer7.norm2.bias", "densenet.features.denseblock3.denselayer7.norm2.running_mean", "densenet.features.denseblock3.denselayer7.norm2.running_var", "densenet.features.denseblock3.denselayer7.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer7.conv2.weight", "densenet.features.denseblock3.denselayer8.norm1.weight", "densenet.features.denseblock3.denselayer8.norm1.bias", "densenet.features.denseblock3.denselayer8.norm1.running_mean", "densenet.features.denseblock3.denselayer8.norm1.running_var", "densenet.features.denseblock3.denselayer8.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer8.conv1.weight", "densenet.features.denseblock3.denselayer8.norm2.weight", "densenet.features.denseblock3.denselayer8.norm2.bias", "densenet.features.denseblock3.denselayer8.norm2.running_mean", "densenet.features.denseblock3.denselayer8.norm2.running_var", "densenet.features.denseblock3.denselayer8.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer8.conv2.weight", "densenet.features.denseblock3.denselayer9.norm1.weight", "densenet.features.denseblock3.denselayer9.norm1.bias", "densenet.features.denseblock3.denselayer9.norm1.running_mean", "densenet.features.denseblock3.denselayer9.norm1.running_var", "densenet.features.denseblock3.denselayer9.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer9.conv1.weight", "densenet.features.denseblock3.denselayer9.norm2.weight", "densenet.features.denseblock3.denselayer9.norm2.bias", "densenet.features.denseblock3.denselayer9.norm2.running_mean", "densenet.features.denseblock3.denselayer9.norm2.running_var", "densenet.features.denseblock3.denselayer9.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer9.conv2.weight", "densenet.features.denseblock3.denselayer10.norm1.weight", "densenet.features.denseblock3.denselayer10.norm1.bias", "densenet.features.denseblock3.denselayer10.norm1.running_mean", "densenet.features.denseblock3.denselayer10.norm1.running_var", "densenet.features.denseblock3.denselayer10.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer10.conv1.weight", "densenet.features.denseblock3.denselayer10.norm2.weight", "densenet.features.denseblock3.denselayer10.norm2.bias", "densenet.features.denseblock3.denselayer10.norm2.running_mean", "densenet.features.denseblock3.denselayer10.norm2.running_var", "densenet.features.denseblock3.denselayer10.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer10.conv2.weight", "densenet.features.denseblock3.denselayer11.norm1.weight", "densenet.features.denseblock3.denselayer11.norm1.bias", "densenet.features.denseblock3.denselayer11.norm1.running_mean", "densenet.features.denseblock3.denselayer11.norm1.running_var", "densenet.features.denseblock3.denselayer11.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer11.conv1.weight", "densenet.features.denseblock3.denselayer11.norm2.weight", "densenet.features.denseblock3.denselayer11.norm2.bias", "densenet.features.denseblock3.denselayer11.norm2.running_mean", "densenet.features.denseblock3.denselayer11.norm2.running_var", "densenet.features.denseblock3.denselayer11.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer11.conv2.weight", "densenet.features.denseblock3.denselayer12.norm1.weight", "densenet.features.denseblock3.denselayer12.norm1.bias", "densenet.features.denseblock3.denselayer12.norm1.running_mean", "densenet.features.denseblock3.denselayer12.norm1.running_var", "densenet.features.denseblock3.denselayer12.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer12.conv1.weight", "densenet.features.denseblock3.denselayer12.norm2.weight", "densenet.features.denseblock3.denselayer12.norm2.bias", "densenet.features.denseblock3.denselayer12.norm2.running_mean", "densenet.features.denseblock3.denselayer12.norm2.running_var", "densenet.features.denseblock3.denselayer12.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer12.conv2.weight", "densenet.features.denseblock3.denselayer13.norm1.weight", "densenet.features.denseblock3.denselayer13.norm1.bias", "densenet.features.denseblock3.denselayer13.norm1.running_mean", "densenet.features.denseblock3.denselayer13.norm1.running_var", "densenet.features.denseblock3.denselayer13.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer13.conv1.weight", "densenet.features.denseblock3.denselayer13.norm2.weight", "densenet.features.denseblock3.denselayer13.norm2.bias", "densenet.features.denseblock3.denselayer13.norm2.running_mean", "densenet.features.denseblock3.denselayer13.norm2.running_var", "densenet.features.denseblock3.denselayer13.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer13.conv2.weight", "densenet.features.denseblock3.denselayer14.norm1.weight", "densenet.features.denseblock3.denselayer14.norm1.bias", "densenet.features.denseblock3.denselayer14.norm1.running_mean", "densenet.features.denseblock3.denselayer14.norm1.running_var", "densenet.features.denseblock3.denselayer14.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer14.conv1.weight", "densenet.features.denseblock3.denselayer14.norm2.weight", "densenet.features.denseblock3.denselayer14.norm2.bias", "densenet.features.denseblock3.denselayer14.norm2.running_mean", "densenet.features.denseblock3.denselayer14.norm2.running_var", "densenet.features.denseblock3.denselayer14.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer14.conv2.weight", "densenet.features.denseblock3.denselayer15.norm1.weight", "densenet.features.denseblock3.denselayer15.norm1.bias", "densenet.features.denseblock3.denselayer15.norm1.running_mean", "densenet.features.denseblock3.denselayer15.norm1.running_var", "densenet.features.denseblock3.denselayer15.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer15.conv1.weight", "densenet.features.denseblock3.denselayer15.norm2.weight", "densenet.features.denseblock3.denselayer15.norm2.bias", "densenet.features.denseblock3.denselayer15.norm2.running_mean", "densenet.features.denseblock3.denselayer15.norm2.running_var", "densenet.features.denseblock3.denselayer15.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer15.conv2.weight", "densenet.features.denseblock3.denselayer16.norm1.weight", "densenet.features.denseblock3.denselayer16.norm1.bias", "densenet.features.denseblock3.denselayer16.norm1.running_mean", "densenet.features.denseblock3.denselayer16.norm1.running_var", "densenet.features.denseblock3.denselayer16.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer16.conv1.weight", "densenet.features.denseblock3.denselayer16.norm2.weight", "densenet.features.denseblock3.denselayer16.norm2.bias", "densenet.features.denseblock3.denselayer16.norm2.running_mean", "densenet.features.denseblock3.denselayer16.norm2.running_var", "densenet.features.denseblock3.denselayer16.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer16.conv2.weight", "densenet.features.denseblock3.denselayer17.norm1.weight", "densenet.features.denseblock3.denselayer17.norm1.bias", "densenet.features.denseblock3.denselayer17.norm1.running_mean", "densenet.features.denseblock3.denselayer17.norm1.running_var", "densenet.features.denseblock3.denselayer17.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer17.conv1.weight", "densenet.features.denseblock3.denselayer17.norm2.weight", "densenet.features.denseblock3.denselayer17.norm2.bias", "densenet.features.denseblock3.denselayer17.norm2.running_mean", "densenet.features.denseblock3.denselayer17.norm2.running_var", "densenet.features.denseblock3.denselayer17.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer17.conv2.weight", "densenet.features.denseblock3.denselayer18.norm1.weight", "densenet.features.denseblock3.denselayer18.norm1.bias", "densenet.features.denseblock3.denselayer18.norm1.running_mean", "densenet.features.denseblock3.denselayer18.norm1.running_var", "densenet.features.denseblock3.denselayer18.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer18.conv1.weight", "densenet.features.denseblock3.denselayer18.norm2.weight", "densenet.features.denseblock3.denselayer18.norm2.bias", "densenet.features.denseblock3.denselayer18.norm2.running_mean", "densenet.features.denseblock3.denselayer18.norm2.running_var", "densenet.features.denseblock3.denselayer18.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer18.conv2.weight", "densenet.features.denseblock3.denselayer19.norm1.weight", "densenet.features.denseblock3.denselayer19.norm1.bias", "densenet.features.denseblock3.denselayer19.norm1.running_mean", "densenet.features.denseblock3.denselayer19.norm1.running_var", "densenet.features.denseblock3.denselayer19.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer19.conv1.weight", "densenet.features.denseblock3.denselayer19.norm2.weight", "densenet.features.denseblock3.denselayer19.norm2.bias", "densenet.features.denseblock3.denselayer19.norm2.running_mean", "densenet.features.denseblock3.denselayer19.norm2.running_var", "densenet.features.denseblock3.denselayer19.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer19.conv2.weight", "densenet.features.denseblock3.denselayer20.norm1.weight", "densenet.features.denseblock3.denselayer20.norm1.bias", "densenet.features.denseblock3.denselayer20.norm1.running_mean", "densenet.features.denseblock3.denselayer20.norm1.running_var", "densenet.features.denseblock3.denselayer20.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer20.conv1.weight", "densenet.features.denseblock3.denselayer20.norm2.weight", "densenet.features.denseblock3.denselayer20.norm2.bias", "densenet.features.denseblock3.denselayer20.norm2.running_mean", "densenet.features.denseblock3.denselayer20.norm2.running_var", "densenet.features.denseblock3.denselayer20.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer20.conv2.weight", "densenet.features.denseblock3.denselayer21.norm1.weight", "densenet.features.denseblock3.denselayer21.norm1.bias", "densenet.features.denseblock3.denselayer21.norm1.running_mean", "densenet.features.denseblock3.denselayer21.norm1.running_var", "densenet.features.denseblock3.denselayer21.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer21.conv1.weight", "densenet.features.denseblock3.denselayer21.norm2.weight", "densenet.features.denseblock3.denselayer21.norm2.bias", "densenet.features.denseblock3.denselayer21.norm2.running_mean", "densenet.features.denseblock3.denselayer21.norm2.running_var", "densenet.features.denseblock3.denselayer21.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer21.conv2.weight", "densenet.features.denseblock3.denselayer22.norm1.weight", "densenet.features.denseblock3.denselayer22.norm1.bias", "densenet.features.denseblock3.denselayer22.norm1.running_mean", "densenet.features.denseblock3.denselayer22.norm1.running_var", "densenet.features.denseblock3.denselayer22.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer22.conv1.weight", "densenet.features.denseblock3.denselayer22.norm2.weight", "densenet.features.denseblock3.denselayer22.norm2.bias", "densenet.features.denseblock3.denselayer22.norm2.running_mean", "densenet.features.denseblock3.denselayer22.norm2.running_var", "densenet.features.denseblock3.denselayer22.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer22.conv2.weight", "densenet.features.denseblock3.denselayer23.norm1.weight", "densenet.features.denseblock3.denselayer23.norm1.bias", "densenet.features.denseblock3.denselayer23.norm1.running_mean", "densenet.features.denseblock3.denselayer23.norm1.running_var", "densenet.features.denseblock3.denselayer23.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer23.conv1.weight", "densenet.features.denseblock3.denselayer23.norm2.weight", "densenet.features.denseblock3.denselayer23.norm2.bias", "densenet.features.denseblock3.denselayer23.norm2.running_mean", "densenet.features.denseblock3.denselayer23.norm2.running_var", "densenet.features.denseblock3.denselayer23.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer23.conv2.weight", "densenet.features.denseblock3.denselayer24.norm1.weight", "densenet.features.denseblock3.denselayer24.norm1.bias", "densenet.features.denseblock3.denselayer24.norm1.running_mean", "densenet.features.denseblock3.denselayer24.norm1.running_var", "densenet.features.denseblock3.denselayer24.norm1.num_batches_tracked", "densenet.features.denseblock3.denselayer24.conv1.weight", "densenet.features.denseblock3.denselayer24.norm2.weight", "densenet.features.denseblock3.denselayer24.norm2.bias", "densenet.features.denseblock3.denselayer24.norm2.running_mean", "densenet.features.denseblock3.denselayer24.norm2.running_var", "densenet.features.denseblock3.denselayer24.norm2.num_batches_tracked", "densenet.features.denseblock3.denselayer24.conv2.weight", "densenet.features.transition3.norm.weight", "densenet.features.transition3.norm.bias", "densenet.features.transition3.norm.running_mean", "densenet.features.transition3.norm.running_var", "densenet.features.transition3.norm.num_batches_tracked", "densenet.features.transition3.conv.weight", "densenet.features.denseblock4.denselayer1.norm1.weight", "densenet.features.denseblock4.denselayer1.norm1.bias", "densenet.features.denseblock4.denselayer1.norm1.running_mean", "densenet.features.denseblock4.denselayer1.norm1.running_var", "densenet.features.denseblock4.denselayer1.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer1.conv1.weight", "densenet.features.denseblock4.denselayer1.norm2.weight", "densenet.features.denseblock4.denselayer1.norm2.bias", "densenet.features.denseblock4.denselayer1.norm2.running_mean", "densenet.features.denseblock4.denselayer1.norm2.running_var", "densenet.features.denseblock4.denselayer1.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer1.conv2.weight", "densenet.features.denseblock4.denselayer2.norm1.weight", "densenet.features.denseblock4.denselayer2.norm1.bias", "densenet.features.denseblock4.denselayer2.norm1.running_mean", "densenet.features.denseblock4.denselayer2.norm1.running_var", "densenet.features.denseblock4.denselayer2.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer2.conv1.weight", "densenet.features.denseblock4.denselayer2.norm2.weight", "densenet.features.denseblock4.denselayer2.norm2.bias", "densenet.features.denseblock4.denselayer2.norm2.running_mean", "densenet.features.denseblock4.denselayer2.norm2.running_var", "densenet.features.denseblock4.denselayer2.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer2.conv2.weight", "densenet.features.denseblock4.denselayer3.norm1.weight", "densenet.features.denseblock4.denselayer3.norm1.bias", "densenet.features.denseblock4.denselayer3.norm1.running_mean", "densenet.features.denseblock4.denselayer3.norm1.running_var", "densenet.features.denseblock4.denselayer3.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer3.conv1.weight", "densenet.features.denseblock4.denselayer3.norm2.weight", "densenet.features.denseblock4.denselayer3.norm2.bias", "densenet.features.denseblock4.denselayer3.norm2.running_mean", "densenet.features.denseblock4.denselayer3.norm2.running_var", "densenet.features.denseblock4.denselayer3.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer3.conv2.weight", "densenet.features.denseblock4.denselayer4.norm1.weight", "densenet.features.denseblock4.denselayer4.norm1.bias", "densenet.features.denseblock4.denselayer4.norm1.running_mean", "densenet.features.denseblock4.denselayer4.norm1.running_var", "densenet.features.denseblock4.denselayer4.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer4.conv1.weight", "densenet.features.denseblock4.denselayer4.norm2.weight", "densenet.features.denseblock4.denselayer4.norm2.bias", "densenet.features.denseblock4.denselayer4.norm2.running_mean", "densenet.features.denseblock4.denselayer4.norm2.running_var", "densenet.features.denseblock4.denselayer4.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer4.conv2.weight", "densenet.features.denseblock4.denselayer5.norm1.weight", "densenet.features.denseblock4.denselayer5.norm1.bias", "densenet.features.denseblock4.denselayer5.norm1.running_mean", "densenet.features.denseblock4.denselayer5.norm1.running_var", "densenet.features.denseblock4.denselayer5.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer5.conv1.weight", "densenet.features.denseblock4.denselayer5.norm2.weight", "densenet.features.denseblock4.denselayer5.norm2.bias", "densenet.features.denseblock4.denselayer5.norm2.running_mean", "densenet.features.denseblock4.denselayer5.norm2.running_var", "densenet.features.denseblock4.denselayer5.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer5.conv2.weight", "densenet.features.denseblock4.denselayer6.norm1.weight", "densenet.features.denseblock4.denselayer6.norm1.bias", "densenet.features.denseblock4.denselayer6.norm1.running_mean", "densenet.features.denseblock4.denselayer6.norm1.running_var", "densenet.features.denseblock4.denselayer6.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer6.conv1.weight", "densenet.features.denseblock4.denselayer6.norm2.weight", "densenet.features.denseblock4.denselayer6.norm2.bias", "densenet.features.denseblock4.denselayer6.norm2.running_mean", "densenet.features.denseblock4.denselayer6.norm2.running_var", "densenet.features.denseblock4.denselayer6.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer6.conv2.weight", "densenet.features.denseblock4.denselayer7.norm1.weight", "densenet.features.denseblock4.denselayer7.norm1.bias", "densenet.features.denseblock4.denselayer7.norm1.running_mean", "densenet.features.denseblock4.denselayer7.norm1.running_var", "densenet.features.denseblock4.denselayer7.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer7.conv1.weight", "densenet.features.denseblock4.denselayer7.norm2.weight", "densenet.features.denseblock4.denselayer7.norm2.bias", "densenet.features.denseblock4.denselayer7.norm2.running_mean", "densenet.features.denseblock4.denselayer7.norm2.running_var", "densenet.features.denseblock4.denselayer7.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer7.conv2.weight", "densenet.features.denseblock4.denselayer8.norm1.weight", "densenet.features.denseblock4.denselayer8.norm1.bias", "densenet.features.denseblock4.denselayer8.norm1.running_mean", "densenet.features.denseblock4.denselayer8.norm1.running_var", "densenet.features.denseblock4.denselayer8.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer8.conv1.weight", "densenet.features.denseblock4.denselayer8.norm2.weight", "densenet.features.denseblock4.denselayer8.norm2.bias", "densenet.features.denseblock4.denselayer8.norm2.running_mean", "densenet.features.denseblock4.denselayer8.norm2.running_var", "densenet.features.denseblock4.denselayer8.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer8.conv2.weight", "densenet.features.denseblock4.denselayer9.norm1.weight", "densenet.features.denseblock4.denselayer9.norm1.bias", "densenet.features.denseblock4.denselayer9.norm1.running_mean", "densenet.features.denseblock4.denselayer9.norm1.running_var", "densenet.features.denseblock4.denselayer9.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer9.conv1.weight", "densenet.features.denseblock4.denselayer9.norm2.weight", "densenet.features.denseblock4.denselayer9.norm2.bias", "densenet.features.denseblock4.denselayer9.norm2.running_mean", "densenet.features.denseblock4.denselayer9.norm2.running_var", "densenet.features.denseblock4.denselayer9.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer9.conv2.weight", "densenet.features.denseblock4.denselayer10.norm1.weight", "densenet.features.denseblock4.denselayer10.norm1.bias", "densenet.features.denseblock4.denselayer10.norm1.running_mean", "densenet.features.denseblock4.denselayer10.norm1.running_var", "densenet.features.denseblock4.denselayer10.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer10.conv1.weight", "densenet.features.denseblock4.denselayer10.norm2.weight", "densenet.features.denseblock4.denselayer10.norm2.bias", "densenet.features.denseblock4.denselayer10.norm2.running_mean", "densenet.features.denseblock4.denselayer10.norm2.running_var", "densenet.features.denseblock4.denselayer10.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer10.conv2.weight", "densenet.features.denseblock4.denselayer11.norm1.weight", "densenet.features.denseblock4.denselayer11.norm1.bias", "densenet.features.denseblock4.denselayer11.norm1.running_mean", "densenet.features.denseblock4.denselayer11.norm1.running_var", "densenet.features.denseblock4.denselayer11.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer11.conv1.weight", "densenet.features.denseblock4.denselayer11.norm2.weight", "densenet.features.denseblock4.denselayer11.norm2.bias", "densenet.features.denseblock4.denselayer11.norm2.running_mean", "densenet.features.denseblock4.denselayer11.norm2.running_var", "densenet.features.denseblock4.denselayer11.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer11.conv2.weight", "densenet.features.denseblock4.denselayer12.norm1.weight", "densenet.features.denseblock4.denselayer12.norm1.bias", "densenet.features.denseblock4.denselayer12.norm1.running_mean", "densenet.features.denseblock4.denselayer12.norm1.running_var", "densenet.features.denseblock4.denselayer12.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer12.conv1.weight", "densenet.features.denseblock4.denselayer12.norm2.weight", "densenet.features.denseblock4.denselayer12.norm2.bias", "densenet.features.denseblock4.denselayer12.norm2.running_mean", "densenet.features.denseblock4.denselayer12.norm2.running_var", "densenet.features.denseblock4.denselayer12.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer12.conv2.weight", "densenet.features.denseblock4.denselayer13.norm1.weight", "densenet.features.denseblock4.denselayer13.norm1.bias", "densenet.features.denseblock4.denselayer13.norm1.running_mean", "densenet.features.denseblock4.denselayer13.norm1.running_var", "densenet.features.denseblock4.denselayer13.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer13.conv1.weight", "densenet.features.denseblock4.denselayer13.norm2.weight", "densenet.features.denseblock4.denselayer13.norm2.bias", "densenet.features.denseblock4.denselayer13.norm2.running_mean", "densenet.features.denseblock4.denselayer13.norm2.running_var", "densenet.features.denseblock4.denselayer13.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer13.conv2.weight", "densenet.features.denseblock4.denselayer14.norm1.weight", "densenet.features.denseblock4.denselayer14.norm1.bias", "densenet.features.denseblock4.denselayer14.norm1.running_mean", "densenet.features.denseblock4.denselayer14.norm1.running_var", "densenet.features.denseblock4.denselayer14.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer14.conv1.weight", "densenet.features.denseblock4.denselayer14.norm2.weight", "densenet.features.denseblock4.denselayer14.norm2.bias", "densenet.features.denseblock4.denselayer14.norm2.running_mean", "densenet.features.denseblock4.denselayer14.norm2.running_var", "densenet.features.denseblock4.denselayer14.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer14.conv2.weight", "densenet.features.denseblock4.denselayer15.norm1.weight", "densenet.features.denseblock4.denselayer15.norm1.bias", "densenet.features.denseblock4.denselayer15.norm1.running_mean", "densenet.features.denseblock4.denselayer15.norm1.running_var", "densenet.features.denseblock4.denselayer15.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer15.conv1.weight", "densenet.features.denseblock4.denselayer15.norm2.weight", "densenet.features.denseblock4.denselayer15.norm2.bias", "densenet.features.denseblock4.denselayer15.norm2.running_mean", "densenet.features.denseblock4.denselayer15.norm2.running_var", "densenet.features.denseblock4.denselayer15.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer15.conv2.weight", "densenet.features.denseblock4.denselayer16.norm1.weight", "densenet.features.denseblock4.denselayer16.norm1.bias", "densenet.features.denseblock4.denselayer16.norm1.running_mean", "densenet.features.denseblock4.denselayer16.norm1.running_var", "densenet.features.denseblock4.denselayer16.norm1.num_batches_tracked", "densenet.features.denseblock4.denselayer16.conv1.weight", "densenet.features.denseblock4.denselayer16.norm2.weight", "densenet.features.denseblock4.denselayer16.norm2.bias", "densenet.features.denseblock4.denselayer16.norm2.running_mean", "densenet.features.denseblock4.denselayer16.norm2.running_var", "densenet.features.denseblock4.denselayer16.norm2.num_batches_tracked", "densenet.features.denseblock4.denselayer16.conv2.weight", "densenet.features.norm5.weight", "densenet.features.norm5.bias", "densenet.features.norm5.running_mean", "densenet.features.norm5.running_var", "densenet.features.norm5.num_batches_tracked", "densenet.classifier.weight", "densenet.classifier.bias", "encoder1.0.denselayer1.norm1.weight", "encoder1.0.denselayer1.norm1.bias", "encoder1.0.denselayer1.norm1.running_mean", "encoder1.0.denselayer1.norm1.running_var", "encoder1.0.denselayer1.norm1.num_batches_tracked", "encoder1.0.denselayer1.conv1.weight", "encoder1.0.denselayer1.norm2.weight", "encoder1.0.denselayer1.norm2.bias", "encoder1.0.denselayer1.norm2.running_mean", "encoder1.0.denselayer1.norm2.running_var", "encoder1.0.denselayer1.norm2.num_batches_tracked", "encoder1.0.denselayer1.conv2.weight", "encoder1.0.denselayer2.norm1.weight", "encoder1.0.denselayer2.norm1.bias", "encoder1.0.denselayer2.norm1.running_mean", "encoder1.0.denselayer2.norm1.running_var", "encoder1.0.denselayer2.norm1.num_batches_tracked", "encoder1.0.denselayer2.conv1.weight", "encoder1.0.denselayer2.norm2.weight", "encoder1.0.denselayer2.norm2.bias", "encoder1.0.denselayer2.norm2.running_mean", "encoder1.0.denselayer2.norm2.running_var", "encoder1.0.denselayer2.norm2.num_batches_tracked", "encoder1.0.denselayer2.conv2.weight", "encoder1.0.denselayer3.norm1.weight", "encoder1.0.denselayer3.norm1.bias", "encoder1.0.denselayer3.norm1.running_mean", "encoder1.0.denselayer3.norm1.running_var", "encoder1.0.denselayer3.norm1.num_batches_tracked", "encoder1.0.denselayer3.conv1.weight", "encoder1.0.denselayer3.norm2.weight", "encoder1.0.denselayer3.norm2.bias", "encoder1.0.denselayer3.norm2.running_mean", "encoder1.0.denselayer3.norm2.running_var", "encoder1.0.denselayer3.norm2.num_batches_tracked", "encoder1.0.denselayer3.conv2.weight", "encoder1.0.denselayer4.norm1.weight", "encoder1.0.denselayer4.norm1.bias", "encoder1.0.denselayer4.norm1.running_mean", "encoder1.0.denselayer4.norm1.running_var", "encoder1.0.denselayer4.norm1.num_batches_tracked", "encoder1.0.denselayer4.conv1.weight", "encoder1.0.denselayer4.norm2.weight", "encoder1.0.denselayer4.norm2.bias", "encoder1.0.denselayer4.norm2.running_mean", "encoder1.0.denselayer4.norm2.running_var", "encoder1.0.denselayer4.norm2.num_batches_tracked", "encoder1.0.denselayer4.conv2.weight", "encoder1.0.denselayer5.norm1.weight", "encoder1.0.denselayer5.norm1.bias", "encoder1.0.denselayer5.norm1.running_mean", "encoder1.0.denselayer5.norm1.running_var", "encoder1.0.denselayer5.norm1.num_batches_tracked", "encoder1.0.denselayer5.conv1.weight", "encoder1.0.denselayer5.norm2.weight", "encoder1.0.denselayer5.norm2.bias", "encoder1.0.denselayer5.norm2.running_mean", "encoder1.0.denselayer5.norm2.running_var", "encoder1.0.denselayer5.norm2.num_batches_tracked", "encoder1.0.denselayer5.conv2.weight", "encoder1.0.denselayer6.norm1.weight", "encoder1.0.denselayer6.norm1.bias", "encoder1.0.denselayer6.norm1.running_mean", "encoder1.0.denselayer6.norm1.running_var", "encoder1.0.denselayer6.norm1.num_batches_tracked", "encoder1.0.denselayer6.conv1.weight", "encoder1.0.denselayer6.norm2.weight", "encoder1.0.denselayer6.norm2.bias", "encoder1.0.denselayer6.norm2.running_mean", "encoder1.0.denselayer6.norm2.running_var", "encoder1.0.denselayer6.norm2.num_batches_tracked", "encoder1.0.denselayer6.conv2.weight", "encoder1.1.norm.weight", "encoder1.1.norm.bias", "encoder1.1.norm.running_mean", "encoder1.1.norm.running_var", "encoder1.1.norm.num_batches_tracked", "encoder1.1.conv.weight", "encoder2.0.denselayer1.norm1.weight", "encoder2.0.denselayer1.norm1.bias", "encoder2.0.denselayer1.norm1.running_mean", "encoder2.0.denselayer1.norm1.running_var", "encoder2.0.denselayer1.norm1.num_batches_tracked", "encoder2.0.denselayer1.conv1.weight", "encoder2.0.denselayer1.norm2.weight", "encoder2.0.denselayer1.norm2.bias", "encoder2.0.denselayer1.norm2.running_mean", "encoder2.0.denselayer1.norm2.running_var", "encoder2.0.denselayer1.norm2.num_batches_tracked", "encoder2.0.denselayer1.conv2.weight", "encoder2.0.denselayer2.norm1.weight", "encoder2.0.denselayer2.norm1.bias", "encoder2.0.denselayer2.norm1.running_mean", "encoder2.0.denselayer2.norm1.running_var", "encoder2.0.denselayer2.norm1.num_batches_tracked", "encoder2.0.denselayer2.conv1.weight", "encoder2.0.denselayer2.norm2.weight", "encoder2.0.denselayer2.norm2.bias", "encoder2.0.denselayer2.norm2.running_mean", "encoder2.0.denselayer2.norm2.running_var", "encoder2.0.denselayer2.norm2.num_batches_tracked", "encoder2.0.denselayer2.conv2.weight", "encoder2.0.denselayer3.norm1.weight", "encoder2.0.denselayer3.norm1.bias", "encoder2.0.denselayer3.norm1.running_mean", "encoder2.0.denselayer3.norm1.running_var", "encoder2.0.denselayer3.norm1.num_batches_tracked", "encoder2.0.denselayer3.conv1.weight", "encoder2.0.denselayer3.norm2.weight", "encoder2.0.denselayer3.norm2.bias", "encoder2.0.denselayer3.norm2.running_mean", "encoder2.0.denselayer3.norm2.running_var", "encoder2.0.denselayer3.norm2.num_batches_tracked", "encoder2.0.denselayer3.conv2.weight", "encoder2.0.denselayer4.norm1.weight", "encoder2.0.denselayer4.norm1.bias", "encoder2.0.denselayer4.norm1.running_mean", "encoder2.0.denselayer4.norm1.running_var", "encoder2.0.denselayer4.norm1.num_batches_tracked", "encoder2.0.denselayer4.conv1.weight", "encoder2.0.denselayer4.norm2.weight", "encoder2.0.denselayer4.norm2.bias", "encoder2.0.denselayer4.norm2.running_mean", "encoder2.0.denselayer4.norm2.running_var", "encoder2.0.denselayer4.norm2.num_batches_tracked", "encoder2.0.denselayer4.conv2.weight", "encoder2.0.denselayer5.norm1.weight", "encoder2.0.denselayer5.norm1.bias", "encoder2.0.denselayer5.norm1.running_mean", "encoder2.0.denselayer5.norm1.running_var", "encoder2.0.denselayer5.norm1.num_batches_tracked", "encoder2.0.denselayer5.conv1.weight", "encoder2.0.denselayer5.norm2.weight", "encoder2.0.denselayer5.norm2.bias", "encoder2.0.denselayer5.norm2.running_mean", "encoder2.0.denselayer5.norm2.running_var", "encoder2.0.denselayer5.norm2.num_batches_tracked", "encoder2.0.denselayer5.conv2.weight", "encoder2.0.denselayer6.norm1.weight", "encoder2.0.denselayer6.norm1.bias", "encoder2.0.denselayer6.norm1.running_mean", "encoder2.0.denselayer6.norm1.running_var", "encoder2.0.denselayer6.norm1.num_batches_tracked", "encoder2.0.denselayer6.conv1.weight", "encoder2.0.denselayer6.norm2.weight", "encoder2.0.denselayer6.norm2.bias", "encoder2.0.denselayer6.norm2.running_mean", "encoder2.0.denselayer6.norm2.running_var", "encoder2.0.denselayer6.norm2.num_batches_tracked", "encoder2.0.denselayer6.conv2.weight", "encoder2.0.denselayer7.norm1.weight", "encoder2.0.denselayer7.norm1.bias", "encoder2.0.denselayer7.norm1.running_mean", "encoder2.0.denselayer7.norm1.running_var", "encoder2.0.denselayer7.norm1.num_batches_tracked", "encoder2.0.denselayer7.conv1.weight", "encoder2.0.denselayer7.norm2.weight", "encoder2.0.denselayer7.norm2.bias", "encoder2.0.denselayer7.norm2.running_mean", "encoder2.0.denselayer7.norm2.running_var", "encoder2.0.denselayer7.norm2.num_batches_tracked", "encoder2.0.denselayer7.conv2.weight", "encoder2.0.denselayer8.norm1.weight", "encoder2.0.denselayer8.norm1.bias", "encoder2.0.denselayer8.norm1.running_mean", "encoder2.0.denselayer8.norm1.running_var", "encoder2.0.denselayer8.norm1.num_batches_tracked", "encoder2.0.denselayer8.conv1.weight", "encoder2.0.denselayer8.norm2.weight", "encoder2.0.denselayer8.norm2.bias", "encoder2.0.denselayer8.norm2.running_mean", "encoder2.0.denselayer8.norm2.running_var", "encoder2.0.denselayer8.norm2.num_batches_tracked", "encoder2.0.denselayer8.conv2.weight", "encoder2.0.denselayer9.norm1.weight", "encoder2.0.denselayer9.norm1.bias", "encoder2.0.denselayer9.norm1.running_mean", "encoder2.0.denselayer9.norm1.running_var", "encoder2.0.denselayer9.norm1.num_batches_tracked", "encoder2.0.denselayer9.conv1.weight", "encoder2.0.denselayer9.norm2.weight", "encoder2.0.denselayer9.norm2.bias", "encoder2.0.denselayer9.norm2.running_mean", "encoder2.0.denselayer9.norm2.running_var", "encoder2.0.denselayer9.norm2.num_batches_tracked", "encoder2.0.denselayer9.conv2.weight", "encoder2.0.denselayer10.norm1.weight", "encoder2.0.denselayer10.norm1.bias", "encoder2.0.denselayer10.norm1.running_mean", "encoder2.0.denselayer10.norm1.running_var", "encoder2.0.denselayer10.norm1.num_batches_tracked", "encoder2.0.denselayer10.conv1.weight", "encoder2.0.denselayer10.norm2.weight", "encoder2.0.denselayer10.norm2.bias", "encoder2.0.denselayer10.norm2.running_mean", "encoder2.0.denselayer10.norm2.running_var", "encoder2.0.denselayer10.norm2.num_batches_tracked", "encoder2.0.denselayer10.conv2.weight", "encoder2.0.denselayer11.norm1.weight", "encoder2.0.denselayer11.norm1.bias", "encoder2.0.denselayer11.norm1.running_mean", "encoder2.0.denselayer11.norm1.running_var", "encoder2.0.denselayer11.norm1.num_batches_tracked", "encoder2.0.denselayer11.conv1.weight", "encoder2.0.denselayer11.norm2.weight", "encoder2.0.denselayer11.norm2.bias", "encoder2.0.denselayer11.norm2.running_mean", "encoder2.0.denselayer11.norm2.running_var", "encoder2.0.denselayer11.norm2.num_batches_tracked", "encoder2.0.denselayer11.conv2.weight", "encoder2.0.denselayer12.norm1.weight", "encoder2.0.denselayer12.norm1.bias", "encoder2.0.denselayer12.norm1.running_mean", "encoder2.0.denselayer12.norm1.running_var", "encoder2.0.denselayer12.norm1.num_batches_tracked", "encoder2.0.denselayer12.conv1.weight", "encoder2.0.denselayer12.norm2.weight", "encoder2.0.denselayer12.norm2.bias", "encoder2.0.denselayer12.norm2.running_mean", "encoder2.0.denselayer12.norm2.running_var", "encoder2.0.denselayer12.norm2.num_batches_tracked", "encoder2.0.denselayer12.conv2.weight", "encoder2.1.norm.weight", "encoder2.1.norm.bias", "encoder2.1.norm.running_mean", "encoder2.1.norm.running_var", "encoder2.1.norm.num_batches_tracked", "encoder2.1.conv.weight", "encoder3.0.denselayer1.norm1.weight", "encoder3.0.denselayer1.norm1.bias", "encoder3.0.denselayer1.norm1.running_mean", "encoder3.0.denselayer1.norm1.running_var", "encoder3.0.denselayer1.norm1.num_batches_tracked", "encoder3.0.denselayer1.conv1.weight", "encoder3.0.denselayer1.norm2.weight", "encoder3.0.denselayer1.norm2.bias", "encoder3.0.denselayer1.norm2.running_mean", "encoder3.0.denselayer1.norm2.running_var", "encoder3.0.denselayer1.norm2.num_batches_tracked", "encoder3.0.denselayer1.conv2.weight", "encoder3.0.denselayer2.norm1.weight", "encoder3.0.denselayer2.norm1.bias", "encoder3.0.denselayer2.norm1.running_mean", "encoder3.0.denselayer2.norm1.running_var", "encoder3.0.denselayer2.norm1.num_batches_tracked", "encoder3.0.denselayer2.conv1.weight", "encoder3.0.denselayer2.norm2.weight", "encoder3.0.denselayer2.norm2.bias", "encoder3.0.denselayer2.norm2.running_mean", "encoder3.0.denselayer2.norm2.running_var", "encoder3.0.denselayer2.norm2.num_batches_tracked", "encoder3.0.denselayer2.conv2.weight", "encoder3.0.denselayer3.norm1.weight", "encoder3.0.denselayer3.norm1.bias", "encoder3.0.denselayer3.norm1.running_mean", "encoder3.0.denselayer3.norm1.running_var", "encoder3.0.denselayer3.norm1.num_batches_tracked", "encoder3.0.denselayer3.conv1.weight", "encoder3.0.denselayer3.norm2.weight", "encoder3.0.denselayer3.norm2.bias", "encoder3.0.denselayer3.norm2.running_mean", "encoder3.0.denselayer3.norm2.running_var", "encoder3.0.denselayer3.norm2.num_batches_tracked", "encoder3.0.denselayer3.conv2.weight", "encoder3.0.denselayer4.norm1.weight", "encoder3.0.denselayer4.norm1.bias", "encoder3.0.denselayer4.norm1.running_mean", "encoder3.0.denselayer4.norm1.running_var", "encoder3.0.denselayer4.norm1.num_batches_tracked", "encoder3.0.denselayer4.conv1.weight", "encoder3.0.denselayer4.norm2.weight", "encoder3.0.denselayer4.norm2.bias", "encoder3.0.denselayer4.norm2.running_mean", "encoder3.0.denselayer4.norm2.running_var", "encoder3.0.denselayer4.norm2.num_batches_tracked", "encoder3.0.denselayer4.conv2.weight", "encoder3.0.denselayer5.norm1.weight", "encoder3.0.denselayer5.norm1.bias", "encoder3.0.denselayer5.norm1.running_mean", "encoder3.0.denselayer5.norm1.running_var", "encoder3.0.denselayer5.norm1.num_batches_tracked", "encoder3.0.denselayer5.conv1.weight", "encoder3.0.denselayer5.norm2.weight", "encoder3.0.denselayer5.norm2.bias", "encoder3.0.denselayer5.norm2.running_mean", "encoder3.0.denselayer5.norm2.running_var", "encoder3.0.denselayer5.norm2.num_batches_tracked", "encoder3.0.denselayer5.conv2.weight", "encoder3.0.denselayer6.norm1.weight", "encoder3.0.denselayer6.norm1.bias", "encoder3.0.denselayer6.norm1.running_mean", "encoder3.0.denselayer6.norm1.running_var", "encoder3.0.denselayer6.norm1.num_batches_tracked", "encoder3.0.denselayer6.conv1.weight", "encoder3.0.denselayer6.norm2.weight", "encoder3.0.denselayer6.norm2.bias", "encoder3.0.denselayer6.norm2.running_mean", "encoder3.0.denselayer6.norm2.running_var", "encoder3.0.denselayer6.norm2.num_batches_tracked", "encoder3.0.denselayer6.conv2.weight", "encoder3.0.denselayer7.norm1.weight", "encoder3.0.denselayer7.norm1.bias", "encoder3.0.denselayer7.norm1.running_mean", "encoder3.0.denselayer7.norm1.running_var", "encoder3.0.denselayer7.norm1.num_batches_tracked", "encoder3.0.denselayer7.conv1.weight", "encoder3.0.denselayer7.norm2.weight", "encoder3.0.denselayer7.norm2.bias", "encoder3.0.denselayer7.norm2.running_mean", "encoder3.0.denselayer7.norm2.running_var", "encoder3.0.denselayer7.norm2.num_batches_tracked", "encoder3.0.denselayer7.conv2.weight", "encoder3.0.denselayer8.norm1.weight", "encoder3.0.denselayer8.norm1.bias", "encoder3.0.denselayer8.norm1.running_mean", "encoder3.0.denselayer8.norm1.running_var", "encoder3.0.denselayer8.norm1.num_batches_tracked", "encoder3.0.denselayer8.conv1.weight", "encoder3.0.denselayer8.norm2.weight", "encoder3.0.denselayer8.norm2.bias", "encoder3.0.denselayer8.norm2.running_mean", "encoder3.0.denselayer8.norm2.running_var", "encoder3.0.denselayer8.norm2.num_batches_tracked", "encoder3.0.denselayer8.conv2.weight", "encoder3.0.denselayer9.norm1.weight", "encoder3.0.denselayer9.norm1.bias", "encoder3.0.denselayer9.norm1.running_mean", "encoder3.0.denselayer9.norm1.running_var", "encoder3.0.denselayer9.norm1.num_batches_tracked", "encoder3.0.denselayer9.conv1.weight", "encoder3.0.denselayer9.norm2.weight", "encoder3.0.denselayer9.norm2.bias", "encoder3.0.denselayer9.norm2.running_mean", "encoder3.0.denselayer9.norm2.running_var", "encoder3.0.denselayer9.norm2.num_batches_tracked", "encoder3.0.denselayer9.conv2.weight", "encoder3.0.denselayer10.norm1.weight", "encoder3.0.denselayer10.norm1.bias", "encoder3.0.denselayer10.norm1.running_mean", "encoder3.0.denselayer10.norm1.running_var", "encoder3.0.denselayer10.norm1.num_batches_tracked", "encoder3.0.denselayer10.conv1.weight", "encoder3.0.denselayer10.norm2.weight", "encoder3.0.denselayer10.norm2.bias", "encoder3.0.denselayer10.norm2.running_mean", "encoder3.0.denselayer10.norm2.running_var", "encoder3.0.denselayer10.norm2.num_batches_tracked", "encoder3.0.denselayer10.conv2.weight", "encoder3.0.denselayer11.norm1.weight", "encoder3.0.denselayer11.norm1.bias", "encoder3.0.denselayer11.norm1.running_mean", "encoder3.0.denselayer11.norm1.running_var", "encoder3.0.denselayer11.norm1.num_batches_tracked", "encoder3.0.denselayer11.conv1.weight", "encoder3.0.denselayer11.norm2.weight", "encoder3.0.denselayer11.norm2.bias", "encoder3.0.denselayer11.norm2.running_mean", "encoder3.0.denselayer11.norm2.running_var", "encoder3.0.denselayer11.norm2.num_batches_tracked", "encoder3.0.denselayer11.conv2.weight", "encoder3.0.denselayer12.norm1.weight", "encoder3.0.denselayer12.norm1.bias", "encoder3.0.denselayer12.norm1.running_mean", "encoder3.0.denselayer12.norm1.running_var", "encoder3.0.denselayer12.norm1.num_batches_tracked", "encoder3.0.denselayer12.conv1.weight", "encoder3.0.denselayer12.norm2.weight", "encoder3.0.denselayer12.norm2.bias", "encoder3.0.denselayer12.norm2.running_mean", "encoder3.0.denselayer12.norm2.running_var", "encoder3.0.denselayer12.norm2.num_batches_tracked", "encoder3.0.denselayer12.conv2.weight", "encoder3.0.denselayer13.norm1.weight", "encoder3.0.denselayer13.norm1.bias", "encoder3.0.denselayer13.norm1.running_mean", "encoder3.0.denselayer13.norm1.running_var", "encoder3.0.denselayer13.norm1.num_batches_tracked", "encoder3.0.denselayer13.conv1.weight", "encoder3.0.denselayer13.norm2.weight", "encoder3.0.denselayer13.norm2.bias", "encoder3.0.denselayer13.norm2.running_mean", "encoder3.0.denselayer13.norm2.running_var", "encoder3.0.denselayer13.norm2.num_batches_tracked", "encoder3.0.denselayer13.conv2.weight", "encoder3.0.denselayer14.norm1.weight", "encoder3.0.denselayer14.norm1.bias", "encoder3.0.denselayer14.norm1.running_mean", "encoder3.0.denselayer14.norm1.running_var", "encoder3.0.denselayer14.norm1.num_batches_tracked", "encoder3.0.denselayer14.conv1.weight", "encoder3.0.denselayer14.norm2.weight", "encoder3.0.denselayer14.norm2.bias", "encoder3.0.denselayer14.norm2.running_mean", "encoder3.0.denselayer14.norm2.running_var", "encoder3.0.denselayer14.norm2.num_batches_tracked", "encoder3.0.denselayer14.conv2.weight", "encoder3.0.denselayer15.norm1.weight", "encoder3.0.denselayer15.norm1.bias", "encoder3.0.denselayer15.norm1.running_mean", "encoder3.0.denselayer15.norm1.running_var", "encoder3.0.denselayer15.norm1.num_batches_tracked", "encoder3.0.denselayer15.conv1.weight", "encoder3.0.denselayer15.norm2.weight", "encoder3.0.denselayer15.norm2.bias", "encoder3.0.denselayer15.norm2.running_mean", "encoder3.0.denselayer15.norm2.running_var", "encoder3.0.denselayer15.norm2.num_batches_tracked", "encoder3.0.denselayer15.conv2.weight", "encoder3.0.denselayer16.norm1.weight", "encoder3.0.denselayer16.norm1.bias", "encoder3.0.denselayer16.norm1.running_mean", "encoder3.0.denselayer16.norm1.running_var", "encoder3.0.denselayer16.norm1.num_batches_tracked", "encoder3.0.denselayer16.conv1.weight", "encoder3.0.denselayer16.norm2.weight", "encoder3.0.denselayer16.norm2.bias", "encoder3.0.denselayer16.norm2.running_mean", "encoder3.0.denselayer16.norm2.running_var", "encoder3.0.denselayer16.norm2.num_batches_tracked", "encoder3.0.denselayer16.conv2.weight", "encoder3.0.denselayer17.norm1.weight", "encoder3.0.denselayer17.norm1.bias", "encoder3.0.denselayer17.norm1.running_mean", "encoder3.0.denselayer17.norm1.running_var", "encoder3.0.denselayer17.norm1.num_batches_tracked", "encoder3.0.denselayer17.conv1.weight", "encoder3.0.denselayer17.norm2.weight", "encoder3.0.denselayer17.norm2.bias", "encoder3.0.denselayer17.norm2.running_mean", "encoder3.0.denselayer17.norm2.running_var", "encoder3.0.denselayer17.norm2.num_batches_tracked", "encoder3.0.denselayer17.conv2.weight", "encoder3.0.denselayer18.norm1.weight", "encoder3.0.denselayer18.norm1.bias", "encoder3.0.denselayer18.norm1.running_mean", "encoder3.0.denselayer18.norm1.running_var", "encoder3.0.denselayer18.norm1.num_batches_tracked", "encoder3.0.denselayer18.conv1.weight", "encoder3.0.denselayer18.norm2.weight", "encoder3.0.denselayer18.norm2.bias", "encoder3.0.denselayer18.norm2.running_mean", "encoder3.0.denselayer18.norm2.running_var", "encoder3.0.denselayer18.norm2.num_batches_tracked", "encoder3.0.denselayer18.conv2.weight", "encoder3.0.denselayer19.norm1.weight", "encoder3.0.denselayer19.norm1.bias", "encoder3.0.denselayer19.norm1.running_mean", "encoder3.0.denselayer19.norm1.running_var", "encoder3.0.denselayer19.norm1.num_batches_tracked", "encoder3.0.denselayer19.conv1.weight", "encoder3.0.denselayer19.norm2.weight", "encoder3.0.denselayer19.norm2.bias", "encoder3.0.denselayer19.norm2.running_mean", "encoder3.0.denselayer19.norm2.running_var", "encoder3.0.denselayer19.norm2.num_batches_tracked", "encoder3.0.denselayer19.conv2.weight", "encoder3.0.denselayer20.norm1.weight", "encoder3.0.denselayer20.norm1.bias", "encoder3.0.denselayer20.norm1.running_mean", "encoder3.0.denselayer20.norm1.running_var", "encoder3.0.denselayer20.norm1.num_batches_tracked", "encoder3.0.denselayer20.conv1.weight", "encoder3.0.denselayer20.norm2.weight", "encoder3.0.denselayer20.norm2.bias", "encoder3.0.denselayer20.norm2.running_mean", "encoder3.0.denselayer20.norm2.running_var", "encoder3.0.denselayer20.norm2.num_batches_tracked", "encoder3.0.denselayer20.conv2.weight", "encoder3.0.denselayer21.norm1.weight", "encoder3.0.denselayer21.norm1.bias", "encoder3.0.denselayer21.norm1.running_mean", "encoder3.0.denselayer21.norm1.running_var", "encoder3.0.denselayer21.norm1.num_batches_tracked", "encoder3.0.denselayer21.conv1.weight", "encoder3.0.denselayer21.norm2.weight", "encoder3.0.denselayer21.norm2.bias", "encoder3.0.denselayer21.norm2.running_mean", "encoder3.0.denselayer21.norm2.running_var", "encoder3.0.denselayer21.norm2.num_batches_tracked", "encoder3.0.denselayer21.conv2.weight", "encoder3.0.denselayer22.norm1.weight", "encoder3.0.denselayer22.norm1.bias", "encoder3.0.denselayer22.norm1.running_mean", "encoder3.0.denselayer22.norm1.running_var", "encoder3.0.denselayer22.norm1.num_batches_tracked", "encoder3.0.denselayer22.conv1.weight", "encoder3.0.denselayer22.norm2.weight", "encoder3.0.denselayer22.norm2.bias", "encoder3.0.denselayer22.norm2.running_mean", "encoder3.0.denselayer22.norm2.running_var", "encoder3.0.denselayer22.norm2.num_batches_tracked", "encoder3.0.denselayer22.conv2.weight", "encoder3.0.denselayer23.norm1.weight", "encoder3.0.denselayer23.norm1.bias", "encoder3.0.denselayer23.norm1.running_mean", "encoder3.0.denselayer23.norm1.running_var", "encoder3.0.denselayer23.norm1.num_batches_tracked", "encoder3.0.denselayer23.conv1.weight", "encoder3.0.denselayer23.norm2.weight", "encoder3.0.denselayer23.norm2.bias", "encoder3.0.denselayer23.norm2.running_mean", "encoder3.0.denselayer23.norm2.running_var", "encoder3.0.denselayer23.norm2.num_batches_tracked", "encoder3.0.denselayer23.conv2.weight", "encoder3.0.denselayer24.norm1.weight", "encoder3.0.denselayer24.norm1.bias", "encoder3.0.denselayer24.norm1.running_mean", "encoder3.0.denselayer24.norm1.running_var", "encoder3.0.denselayer24.norm1.num_batches_tracked", "encoder3.0.denselayer24.conv1.weight", "encoder3.0.denselayer24.norm2.weight", "encoder3.0.denselayer24.norm2.bias", "encoder3.0.denselayer24.norm2.running_mean", "encoder3.0.denselayer24.norm2.running_var", "encoder3.0.denselayer24.norm2.num_batches_tracked", "encoder3.0.denselayer24.conv2.weight", "encoder3.1.norm.weight", "encoder3.1.norm.bias", "encoder3.1.norm.running_mean", "encoder3.1.norm.running_var", "encoder3.1.norm.num_batches_tracked", "encoder3.1.conv.weight", "encoder4.0.denselayer1.norm1.weight", "encoder4.0.denselayer1.norm1.bias", "encoder4.0.denselayer1.norm1.running_mean", "encoder4.0.denselayer1.norm1.running_var", "encoder4.0.denselayer1.norm1.num_batches_tracked", "encoder4.0.denselayer1.conv1.weight", "encoder4.0.denselayer1.norm2.weight", "encoder4.0.denselayer1.norm2.bias", "encoder4.0.denselayer1.norm2.running_mean", "encoder4.0.denselayer1.norm2.running_var", "encoder4.0.denselayer1.norm2.num_batches_tracked", "encoder4.0.denselayer1.conv2.weight", "encoder4.0.denselayer2.norm1.weight", "encoder4.0.denselayer2.norm1.bias", "encoder4.0.denselayer2.norm1.running_mean", "encoder4.0.denselayer2.norm1.running_var", "encoder4.0.denselayer2.norm1.num_batches_tracked", "encoder4.0.denselayer2.conv1.weight", "encoder4.0.denselayer2.norm2.weight", "encoder4.0.denselayer2.norm2.bias", "encoder4.0.denselayer2.norm2.running_mean", "encoder4.0.denselayer2.norm2.running_var", "encoder4.0.denselayer2.norm2.num_batches_tracked", "encoder4.0.denselayer2.conv2.weight", "encoder4.0.denselayer3.norm1.weight", "encoder4.0.denselayer3.norm1.bias", "encoder4.0.denselayer3.norm1.running_mean", "encoder4.0.denselayer3.norm1.running_var", "encoder4.0.denselayer3.norm1.num_batches_tracked", "encoder4.0.denselayer3.conv1.weight", "encoder4.0.denselayer3.norm2.weight", "encoder4.0.denselayer3.norm2.bias", "encoder4.0.denselayer3.norm2.running_mean", "encoder4.0.denselayer3.norm2.running_var", "encoder4.0.denselayer3.norm2.num_batches_tracked", "encoder4.0.denselayer3.conv2.weight", "encoder4.0.denselayer4.norm1.weight", "encoder4.0.denselayer4.norm1.bias", "encoder4.0.denselayer4.norm1.running_mean", "encoder4.0.denselayer4.norm1.running_var", "encoder4.0.denselayer4.norm1.num_batches_tracked", "encoder4.0.denselayer4.conv1.weight", "encoder4.0.denselayer4.norm2.weight", "encoder4.0.denselayer4.norm2.bias", "encoder4.0.denselayer4.norm2.running_mean", "encoder4.0.denselayer4.norm2.running_var", "encoder4.0.denselayer4.norm2.num_batches_tracked", "encoder4.0.denselayer4.conv2.weight", "encoder4.0.denselayer5.norm1.weight", "encoder4.0.denselayer5.norm1.bias", "encoder4.0.denselayer5.norm1.running_mean", "encoder4.0.denselayer5.norm1.running_var", "encoder4.0.denselayer5.norm1.num_batches_tracked", "encoder4.0.denselayer5.conv1.weight", "encoder4.0.denselayer5.norm2.weight", "encoder4.0.denselayer5.norm2.bias", "encoder4.0.denselayer5.norm2.running_mean", "encoder4.0.denselayer5.norm2.running_var", "encoder4.0.denselayer5.norm2.num_batches_tracked", "encoder4.0.denselayer5.conv2.weight", "encoder4.0.denselayer6.norm1.weight", "encoder4.0.denselayer6.norm1.bias", "encoder4.0.denselayer6.norm1.running_mean", "encoder4.0.denselayer6.norm1.running_var", "encoder4.0.denselayer6.norm1.num_batches_tracked", "encoder4.0.denselayer6.conv1.weight", "encoder4.0.denselayer6.norm2.weight", "encoder4.0.denselayer6.norm2.bias", "encoder4.0.denselayer6.norm2.running_mean", "encoder4.0.denselayer6.norm2.running_var", "encoder4.0.denselayer6.norm2.num_batches_tracked", "encoder4.0.denselayer6.conv2.weight", "encoder4.0.denselayer7.norm1.weight", "encoder4.0.denselayer7.norm1.bias", "encoder4.0.denselayer7.norm1.running_mean", "encoder4.0.denselayer7.norm1.running_var", "encoder4.0.denselayer7.norm1.num_batches_tracked", "encoder4.0.denselayer7.conv1.weight", "encoder4.0.denselayer7.norm2.weight", "encoder4.0.denselayer7.norm2.bias", "encoder4.0.denselayer7.norm2.running_mean", "encoder4.0.denselayer7.norm2.running_var", "encoder4.0.denselayer7.norm2.num_batches_tracked", "encoder4.0.denselayer7.conv2.weight", "encoder4.0.denselayer8.norm1.weight", "encoder4.0.denselayer8.norm1.bias", "encoder4.0.denselayer8.norm1.running_mean", "encoder4.0.denselayer8.norm1.running_var", "encoder4.0.denselayer8.norm1.num_batches_tracked", "encoder4.0.denselayer8.conv1.weight", "encoder4.0.denselayer8.norm2.weight", "encoder4.0.denselayer8.norm2.bias", "encoder4.0.denselayer8.norm2.running_mean", "encoder4.0.denselayer8.norm2.running_var", "encoder4.0.denselayer8.norm2.num_batches_tracked", "encoder4.0.denselayer8.conv2.weight", "encoder4.0.denselayer9.norm1.weight", "encoder4.0.denselayer9.norm1.bias", "encoder4.0.denselayer9.norm1.running_mean", "encoder4.0.denselayer9.norm1.running_var", "encoder4.0.denselayer9.norm1.num_batches_tracked", "encoder4.0.denselayer9.conv1.weight", "encoder4.0.denselayer9.norm2.weight", "encoder4.0.denselayer9.norm2.bias", "encoder4.0.denselayer9.norm2.running_mean", "encoder4.0.denselayer9.norm2.running_var", "encoder4.0.denselayer9.norm2.num_batches_tracked", "encoder4.0.denselayer9.conv2.weight", "encoder4.0.denselayer10.norm1.weight", "encoder4.0.denselayer10.norm1.bias", "encoder4.0.denselayer10.norm1.running_mean", "encoder4.0.denselayer10.norm1.running_var", "encoder4.0.denselayer10.norm1.num_batches_tracked", "encoder4.0.denselayer10.conv1.weight", "encoder4.0.denselayer10.norm2.weight", "encoder4.0.denselayer10.norm2.bias", "encoder4.0.denselayer10.norm2.running_mean", "encoder4.0.denselayer10.norm2.running_var", "encoder4.0.denselayer10.norm2.num_batches_tracked", "encoder4.0.denselayer10.conv2.weight", "encoder4.0.denselayer11.norm1.weight", "encoder4.0.denselayer11.norm1.bias", "encoder4.0.denselayer11.norm1.running_mean", "encoder4.0.denselayer11.norm1.running_var", "encoder4.0.denselayer11.norm1.num_batches_tracked", "encoder4.0.denselayer11.conv1.weight", "encoder4.0.denselayer11.norm2.weight", "encoder4.0.denselayer11.norm2.bias", "encoder4.0.denselayer11.norm2.running_mean", "encoder4.0.denselayer11.norm2.running_var", "encoder4.0.denselayer11.norm2.num_batches_tracked", "encoder4.0.denselayer11.conv2.weight", "encoder4.0.denselayer12.norm1.weight", "encoder4.0.denselayer12.norm1.bias", "encoder4.0.denselayer12.norm1.running_mean", "encoder4.0.denselayer12.norm1.running_var", "encoder4.0.denselayer12.norm1.num_batches_tracked", "encoder4.0.denselayer12.conv1.weight", "encoder4.0.denselayer12.norm2.weight", "encoder4.0.denselayer12.norm2.bias", "encoder4.0.denselayer12.norm2.running_mean", "encoder4.0.denselayer12.norm2.running_var", "encoder4.0.denselayer12.norm2.num_batches_tracked", "encoder4.0.denselayer12.conv2.weight", "encoder4.0.denselayer13.norm1.weight", "encoder4.0.denselayer13.norm1.bias", "encoder4.0.denselayer13.norm1.running_mean", "encoder4.0.denselayer13.norm1.running_var", "encoder4.0.denselayer13.norm1.num_batches_tracked", "encoder4.0.denselayer13.conv1.weight", "encoder4.0.denselayer13.norm2.weight", "encoder4.0.denselayer13.norm2.bias", "encoder4.0.denselayer13.norm2.running_mean", "encoder4.0.denselayer13.norm2.running_var", "encoder4.0.denselayer13.norm2.num_batches_tracked", "encoder4.0.denselayer13.conv2.weight", "encoder4.0.denselayer14.norm1.weight", "encoder4.0.denselayer14.norm1.bias", "encoder4.0.denselayer14.norm1.running_mean", "encoder4.0.denselayer14.norm1.running_var", "encoder4.0.denselayer14.norm1.num_batches_tracked", "encoder4.0.denselayer14.conv1.weight", "encoder4.0.denselayer14.norm2.weight", "encoder4.0.denselayer14.norm2.bias", "encoder4.0.denselayer14.norm2.running_mean", "encoder4.0.denselayer14.norm2.running_var", "encoder4.0.denselayer14.norm2.num_batches_tracked", "encoder4.0.denselayer14.conv2.weight", "encoder4.0.denselayer15.norm1.weight", "encoder4.0.denselayer15.norm1.bias", "encoder4.0.denselayer15.norm1.running_mean", "encoder4.0.denselayer15.norm1.running_var", "encoder4.0.denselayer15.norm1.num_batches_tracked", "encoder4.0.denselayer15.conv1.weight", "encoder4.0.denselayer15.norm2.weight", "encoder4.0.denselayer15.norm2.bias", "encoder4.0.denselayer15.norm2.running_mean", "encoder4.0.denselayer15.norm2.running_var", "encoder4.0.denselayer15.norm2.num_batches_tracked", "encoder4.0.denselayer15.conv2.weight", "encoder4.0.denselayer16.norm1.weight", "encoder4.0.denselayer16.norm1.bias", "encoder4.0.denselayer16.norm1.running_mean", "encoder4.0.denselayer16.norm1.running_var", "encoder4.0.denselayer16.norm1.num_batches_tracked", "encoder4.0.denselayer16.conv1.weight", "encoder4.0.denselayer16.norm2.weight", "encoder4.0.denselayer16.norm2.bias", "encoder4.0.denselayer16.norm2.running_mean", "encoder4.0.denselayer16.norm2.running_var", "encoder4.0.denselayer16.norm2.num_batches_tracked", "encoder4.0.denselayer16.conv2.weight", "encoder4.1.weight", "encoder4.1.bias", "encoder4.2.weight", "encoder4.2.bias", "encoder4.2.running_mean", "encoder4.2.running_var", "encoder4.2.num_batches_tracked", "decoder0_g.1.weight", "decoder0_g.1.bias", "decoder0_g.1.running_mean", "decoder0_g.1.running_var", "decoder0_g.1.num_batches_tracked", "decoder0_g.3.weight", "decoder0_g.3.bias", "decoder0_g.4.weight", "decoder0_g.4.bias", "decoder0_g.4.running_mean", "decoder0_g.4.running_var", "decoder0_g.4.num_batches_tracked", "decoder0_g.6.weight", "decoder0_g.6.bias", "decoder0_f.1.weight", "decoder0_f.1.bias", "decoder0_f.1.running_mean", "decoder0_f.1.running_var", "decoder0_f.1.num_batches_tracked", "decoder0_f.3.weight", "decoder0_f.3.bias", "decoder0_f.4.weight", "decoder0_f.4.bias", "decoder0_f.4.running_mean", "decoder0_f.4.running_var", "decoder0_f.4.num_batches_tracked", "decoder0_f.6.weight", "decoder0_f.6.bias". 
	size mismatch for encoder0.0.weight: copying a param with shape torch.Size([64, 3, 7, 7]) from checkpoint, the shape in current model is torch.Size([64, 3, 3, 3]).
	size mismatch for decoder3_f.0.weight: copying a param with shape torch.Size([256, 768, 3, 3]) from checkpoint, the shape in current model is torch.Size([256, 512, 3, 3]).
	size mismatch for decoder2_f.0.weight: copying a param with shape torch.Size([128, 384, 3, 3]) from checkpoint, the shape in current model is torch.Size([128, 256, 3, 3]).
	size mismatch for decoder1_f.0.weight: copying a param with shape torch.Size([64, 192, 3, 3]) from checkpoint, the shape in current model is torch.Size([64, 128, 3, 3]).
	size mismatch for decoder0_g.0.weight: copying a param with shape torch.Size([64, 128, 3, 3]) from checkpoint, the shape in current model is torch.Size([3, 64, 3, 3]).
	size mismatch for decoder0_g.0.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([3]).
	size mismatch for decoder0_f.0.weight: copying a param with shape torch.Size([64, 128, 3, 3]) from checkpoint, the shape in current model is torch.Size([1, 64, 3, 3]).
	size mismatch for decoder0_f.0.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([1]).

I'm running the code in Windows using Anaconda 4.10.3. Would you be able to help me out with this? It feels like the model files and code are out of sync with each other.

question about article

Ref Hatt[9] didn't share their code. How did you compare results in figure.6 ?

Will you publish AM-2K dataset?

This paper published about 9 month ago and you already cite AM2K dataset in next paper (https://arxiv.org/pdf/2107.07235v1.pdf), but dataset is still not published
:(

question about training result....

firstly thanks for your greate job~~~~
and we try to use this method on human portrait matting, recently we implement the training code according to your paper and inference code ...
we find the result is a little weird,especially the focus_sigmoid...

The pictures are as follows:
【[1,ori_img 2，focus_sigmoid，3，trimap_mask],
[4，fg_mask，5，focus_sigmoid*trimap_mask 6，focus_sigmoid*trimap_mask+fg_mask]】

so we want to consult you,is there something to be noted in the training phase? or the focus_sigmoid is normal cause of not enough training time? or what something else?

looking forward to your reply~~~~

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.