yangyuke001 / fiiqa-pytorch Goto Github PK

Hi, there is the scripts for illumination transfer? I want to generate training data by myself.

您好，运行 python train.py 提示错误：NameError: name 'AGLoss' is not defined

你好，请问是不是模型预测出的分数越高，表示这个图片质量越好？还有图片标签从0-199，是不是表示图片从欠曝-->过曝-->正常？？？

[root@e31325cc657e FIIQA-PyTorch]# python3 test.py
Traceback (most recent call last):
File "test.py", line 48, in
net.load_state_dict(checkpoint['net'])
File "/usr/local/lib64/python3.6/site-packages/torch/nn/modules/module.py", line 1045, in load_state_dict
self.class.name, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for ShuffleNetV2:
size mismatch for classifier.0.weight: copying a param with shape torch.Size([200, 1024]) from checkpoint, the shape in current model is torch.Size([160, 1024]).
size mismatch for classifier.0.bias: copying a param with shape torch.Size([200]) from checkpoint, the shape in current model is torch.Size([160]).

facePath = imagePath + "/" + img
face = Image.open(facePath)
imgblob = dataTransforms(face).unsqueeze(0)
imgblob = Variable(imgblob)
torch.no_grad()
predict = F.softmax(net(imgblob), dim=1)

predict是一个长度为2的一维张量，正常情况下是长度为1的一维张量。
这是我的邮箱，[email protected]。需要的话，我可以将这样的图片发给您。

你好，我请教一下，test_convert_to_caffe这个目录下的文件是用来实现pytorch转caffe的吗？
我运行python shufflenetv2_to_caffe.py的时候报错（python3.6.10）

Traceback (most recent call last):
  File "shufflenetv2_to_caffe.py", line 16, in <module>
    import converter
ModuleNotFoundError: No module named 'converter'

用pip install converter安装converter后，出现交互式提示：

 This is the metric converter 
What would you like to convert ?
You can either convert a measurement, weight or volume 
>

这里我应该怎么选择？

您好，您方便提供illumination transfer 工具吗？

I use the code to do my validation. I want to evaluate the illumination quality of face images in Yaleb DATABASE. The bug is
Traceback (most recent call last):
File "/home/./桌面/FIIQA-PyTorch-master/test2.py", line 65, in
imgblob = dataTransforms(face).unsqueeze(0)
File "/home/./anaconda3/envs/pytorch/lib/python3.7/site-packages/torchvision/transforms/transforms.py", line 67, in call
img = t(img)
File "/home/./anaconda3/envs/pytorch/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in call_impl
result = self.forward(*input, **kwargs)
File "/home/./anaconda3/envs/pytorch/lib/python3.7/site-packages/torchvision/transforms/transforms.py", line 226, in forward
return F.normalize(tensor, self.mean, self.std, self.inplace)
File "/home/./anaconda3/envs/pytorch/lib/python3.7/site-packages/torchvision/transforms/functional.py", line 284, in normalize
tensor.sub(mean).div_(std)
RuntimeError: output with shape [1, 228, 200] doesn't match the broadcast shape [3, 228, 200]

Do you have any solutions ? thanks.

Hi,
I started training from scratch without using the pre_trained model 97_160_2.pth
My training results were poor

I use FIIQALoss(),and optimizer is AdaBound.
Even if I loaded the pre_trained model, I didn't get better results, why?
Thanks