deepcam-cn / facequality Goto Github PK

Is there any pretrained model that we can download (not baidu)

Thanks for sharing this very interesting work. Right now, no license has been included in the repo, meaning all rights are reserved (https://opensource.stackexchange.com/questions/1720/what-can-i-assume-if-a-publicly-published-project-has-no-license), and it cannot be used by others.

Could you perhaps consider including some form of open-source license?

Hi,

I try to conver to onnx but not succesfull ? the bacbone model only results 512 but it should more

Best

Hi, I've downloaded your pretrained model from google drive, but when i run test_quality.py, it returns mentioned error. what should I do to solve it? should I train it from scratch? thanks in advance

File "train_quality.py", line 179, in
train()
File "train_quality.py", line 127, in train
confidence = QUALITY(fc)
File "/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py", line 722, in _call_impl
result = self.forward(*input, **kwargs)
File "/opt/conda/lib/python3.6/site-packages/torch/nn/parallel/data_parallel.py", line 155, in forward
outputs = self.parallel_apply(replicas, inputs, kwargs)
File "/opt/conda/lib/python3.6/site-packages/torch/nn/parallel/data_parallel.py", line 165, in parallel_apply
return parallel_apply(replicas, inputs, kwargs, self.device_ids[:len(replicas)])
File "/opt/conda/lib/python3.6/site-packages/torch/nn/parallel/parallel_apply.py", line 85, in parallel_apply
output.reraise()
File "/opt/conda/lib/python3.6/site-packages/torch/_utils.py", line 395, in reraise
raise self.exc_type(msg)
RuntimeError: Caught RuntimeError in replica 0 on device 0.

File "/opt/conda/lib/python3.6/site-packages/torch/nn/parallel/parallel_apply.py", line 60, in _worker
output = module(*input, **kwargs)
File "/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py", line 722, in _call_impl
result = self.forward(*input, **kwargs)
File "/tmp/facequality/models/model_resnet.py", line 110, in forward
x = self.qualtiy(x)
File "/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py", line 722, in _call_impl
result = self.forward(*input, **kwargs)
File "/opt/conda/lib/python3.6/site-packages/torch/nn/modules/container.py", line 117, in forward
input = module(input)
File "/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py", line 722, in _call_impl
result = self.forward(*input, **kwargs)
File "/opt/conda/lib/python3.6/site-packages/torch/nn/modules/linear.py", line 91, in forward
return F.linear(input, self.weight, self.bias)
File "/opt/conda/lib/python3.6/site-packages/torch/nn/functional.py", line 1676, in linear
output = input.matmul(weight.t())
RuntimeError: mat1 dim 1 must match mat2 dim 0

https://github.com/deepinsight/insightface/blob/master/recognition/common/rec2image.py

你好，我在你给的连接里下载了faces_webface_112x112数据集。但是你给的rec2image.py连接没有了。现在不能生成训练文件。您能再提供一下rec2image.py吗？

Thank you for perfect repository <3
I see that you provided code to training model Arcface using Resnet backbone?
So I want to train with MobileFacenet backbone.
Can you help me resolve this problem?
Thank you <3

作者您好，如题。希望能提供代码，方便复现工作，蟹蟹～

Hi, Thanks for your amazing work.

However, I have a question why does this implementation use focal loss instead of the loss function below that we see in the paper?

In the paper, you said that it would be decayed by 10 after 30, 60, 90 epoch for total of 100 epochs. But in the code, I saw that you were using CosineAnnealingRate, which doesn't have the effect as the above.
And also, I saw that u pass T_max hard-code 10 epochs (10 * len(train_loader)) -> is this intentional? Cause this would make the LR varies in a cyclical way.
Thank you for reading and answering.

I see the config in step 2 is:
...
BACKBONE_RESUME_ROOT = './backbone_resume.pth'
HEAD_RESUME_ROOT = './head_resume.pth'
TRAIN_FILES = './dataset/face_train_ms1mv2.txt'
BACKBONE_LR = 0.05
PRETRAINED_BACKBONE = ''
PRETRAINED_QUALITY = ''
...
So where can i get the backbone_resume.pth and head_resume.pth
And where can i get pretrained_backbone_resume.pth and pretrained_qulity_resume.pth in step 3?
...
BACKBONE_RESUME_ROOT = ''
HEAD_RESUME_ROOT = ''
TRAIN_FILES = './dataset/face_train_ms1mv2.txt'
BACKBONE_LR = 0.05
PRETRAINED_BACKBONE = ''
PRETRAINED_QUALITY = ''
PRETRAINED_BACKBONE = 'pretrained_backbone_resume.pth'
PRETRAINED_QUALITY = 'pretrained_qulity_resume.pth'
...

Hello,I am very interested in your work.and I want to ask some questions, why is the implementation of the code inconsistent with the description in the paper, such as training parameters, learning rate strategy, etc.