Comments (8)
我试过了!昨天刚试过了
MIT-Adobe FiveK 数据集, ExpertC作为标签, 0-4499 做训练集,4500-5000 做验证集,训练了 95 epoch 得以下结果:
opencv_psnr : 15.675621419654806
mean_psnr : 28.08877043728922
sklearn_ssim : 0.7421423059845486
虽然客观评测的数据不是很好看,但主观上的增强我觉得是众多方法中较好的!EnlightenGAN 在低光照这方面做得很好,并没有失真啊之类的情况,给几个例子吧:
Bad Images with low light
Enhanced with EnlightenGAN which is trained with Adobe FiveK dataset
from enlightengan.
有没有人试过在别的数据集上训练模型呢?需要注意什么吗,谢谢
from enlightengan.
些许改动
-
script.py 那里要改数据集的路径,分别命名为 testA 和 testB
-
要注意数据集中短边最小的长度,因为有个 fineSize 是用来切割图像的,在 script 中修改就可以了
-
注意测试的图像不能太大,会 OOM 的,我试过了,2000 * 1500 分辨率的图像还是可以的,11G 1080ti。可惜测试这块,作者没有实现多块卡上 Inference,如果可以在低分辨率下增强,作用到高分辨率图像上,就好比 HDRnet 那种就好了
from enlightengan.
请问是否需要修改gpu_ids参数呢
from enlightengan.
@hermosayhl
Can you share the dataset you have used (A and B)?
from enlightengan.
@hermosayhl
请问在cpu上能跑吗?
from enlightengan.
你是使用自己的低亮度数据集和作者的提供的trainB吗?可以说说您怎么构建您的数据集的吗?我用bdd-100k数据集作为trainA,作者提供的trainB训练,效果还不是很好,可以介绍您除了修改数据集路径还修改哪些参数吗? @yueruchen @hermosayhl
from enlightengan.
@zhangxiaopang88 Could you please provide some examples so that I can help debug? And maybe you could try to tune -vgg
to adjust your own dataset
from enlightengan.
Related Issues (20)
- why does image_numpy need to add one,and divide 2,finally multipy 255 HOT 2
- Code for calculate NIQE HOT 1
- Problems about how to fine-tuning HOT 3
- why loss is nan
- predict.py: error: the following arguments are required: --dataroot HOT 2
- UserWarning: nn.functional.upsample is deprecated. Use nn.functional.interpolate instead.
- multiprocessing error HOT 2
- Could it be posible create low-light image? HOT 1
- 请问有关于纯粹的EnlightenGAN这篇论文的代码吗?不包含其他对比算法的。 HOT 1
- train issues HOT 1
- where to place the final_dataset folder? HOT 1
- 请问在哪里修改生成图像的名称 HOT 1
- error HOT 1
- TypeError: load() missing 1 required positional argument: 'Loader' HOT 1
- The difference between D_A and D_P loss
- 训练时判别器网络报错
- 下载模型 HOT 1
- Pre-Processing for Improving Classification
- Predict the .mp4 HOT 1
- 预测时只能调用一张显卡
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.
from enlightengan.