RuntimeError: The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 0

[WXY-Belief]

你好，我想问一下，为什么会出现这个问题呀?
Hello, I'd like to ask, why is this problem?

Traceback (most recent call last):
File "easy_use.py", line 97, in
generate_results(args, device)
File "easy_use.py", line 66, in generate_results
img = transform(img)
File "/disk/sdb/public_user/anaconda3/envs/wxy_env/lib/python3.8/site-packages/torchvision/transforms/transforms.py", line 67, in call
img = t(img)
File "/disk/sdb/public_user/anaconda3/envs/wxy_env/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in call_impl
result = self.forward(*input, **kwargs)
File "/disk/sdb/public_user/anaconda3/envs/wxy_env/lib/python3.8/site-packages/torchvision/transforms/transforms.py", line 226, in forward
return F.normalize(tensor, self.mean, self.std, self.inplace)
File "/disk/sdb/public_user/anaconda3/envs/wxy_env/lib/python3.8/site-packages/torchvision/transforms/functional.py", line 284, in normalize
tensor.sub(mean).div_(std)
RuntimeError: The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 0

[StoryMY]

It seems that the channel of img is 4 (RGBA), which is expected to be 3 (RGB).
You can print(np.array(img).shape) before "img = transform(img)" to check it.
The correct shape is expected to be (256, 256, 3).

If your result is (256, 256, 4), you need to convert it to RGB first.
Maybe like this: "img = Image.open(os.path.join(args.input_dir, img_name)).convert('RGB')" (Line 64)

[WXY-Belief]

Well, thank you very much.I found a solution on Baidu.But now there is another problem. All my documents are shown in the attachment. The program ran for a long time without results. I don't know if this is normal.

…

------------------ 原始邮件 ------------------ 发件人: "StoryMY/take-off-eyeglasses" ***@***.***>; 发送时间: 2022年5月10日(星期二) 下午5:22 ***@***.***>; ***@***.******@***.***>; 主题: Re: [StoryMY/take-off-eyeglasses] RuntimeError: The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 0 (Issue #1) It seems that the channel of img is 4 (RGBA), which is expected to be 3 (RGB). You can print(img.shape) before "img = transform(img)" to check it. The correct shape is expected to be torch.Size([3, 256, 256]). If your result is torch.Size([4, 256, 256]), you need to convert it to RGB first. Maybe like this: "img = Image.open(os.path.join(args.input_dir, img_name)).convert('RGB')" (Line 64) — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[WXY-Belief]

Sorry, there is something wrong with my configuration environment

…

------------------ 原始邮件 ------------------ 发件人: "StoryMY/take-off-eyeglasses" ***@***.***>; 发送时间: 2022年5月10日(星期二) 下午5:22 ***@***.***>; ***@***.******@***.***>; 主题: Re: [StoryMY/take-off-eyeglasses] RuntimeError: The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 0 (Issue #1) It seems that the channel of img is 4 (RGBA), which is expected to be 3 (RGB). You can print(img.shape) before "img = transform(img)" to check it. The correct shape is expected to be torch.Size([3, 256, 256]). If your result is torch.Size([4, 256, 256]), you need to convert it to RGB first. Maybe like this: "img = Image.open(os.path.join(args.input_dir, img_name)).convert('RGB')" (Line 64) — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>