jingzhang617 / ucnet Goto Github PK

Hi!Thank you for the code, it is easy to run in my machine.
However,I still have several problems after reading the code, could you please give me some hint?

1. I want to know how to get multiple results generated from one image.
2. which part in the code is corresponding to the saliency consensus module in the paper.

Traceback (most recent call last):
File "train.py", line 136, in
pred_post, pred_prior, lattent_loss, depth_pred_post, depth_pred_prior = generator.forward(images,depths,gts)
File "/home/paramshavakmahesha/Documents/TRY4/UCNet/model/ResNet_models.py", line 159, in forward
lattent_loss = torch.mean(self.kl_divergence(self.posterior, self.prior))
File "/home/paramshavakmahesha/Documents/TRY4/UCNet/model/ResNet_models.py", line 146, in kl_divergence
kl_div = kl.kl_divergence(posterior_latent_space, prior_latent_space)
File "/home/paramshavakmahesha/anaconda3/envs/picanet_3.5/lib/python3.5/site-packages/torch/distributions/kl.py", line 161, in kl_divergence
raise NotImplementedError
NotImplementedError

As shown in competing_results_show.png, there are ours1 and ours2. However, only one result generated from one image in test.py. I want to know how to get multiple results generated from one image.

Hello,I have problem with this,i have not found DepthCorrectionNet in your code

python train.py
Traceback (most recent call last):
File "train.py", line 8, in
from model.ResNet_models import Generator
File "/home/ali/UCNet-master/model/ResNet_models.py", line 5, in
from ResNet import B2_ResNet
ModuleNotFoundError: No module named 'ResNet'

Hello,

Have you encountered this error during training ?

Traceback (most recent call last):
File "train.py", line 136, in
pred_post, pred_prior, latent_loss, depth_pred_post, depth_pred_prior = generator.forward(images,depths,gts)
File "/content/UCNet/model/ResNet_models.py", line 157, in forward
self.posterior, muxy, logvarxy = self.xy_encoder(torch.cat((x,depth,y),1))
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/content/UCNet/model/ResNet_models.py", line 125, in forward
dist = Independent(Normal(loc=mu, scale=torch.exp(logvar)), 1)
File "/usr/local/lib/python3.7/dist-packages/torch/distributions/normal.py", line 50, in init
super(Normal, self).init(batch_shape, validate_args=validate_args)
File "/usr/local/lib/python3.7/dist-packages/torch/distributions/distribution.py", line 56, in init
f"Expected parameter {param} "
ValueError: Expected parameter loc (Tensor of shape (10, 3)) of distribution Normal(loc: torch.Size([10, 3]), scale: torch.Size([10, 3])) to satisfy the constraint Real(), but found invalid values:
tensor([[nan, nan, nan],
[nan, nan, nan],
[nan, nan, nan],
[nan, nan, nan],
[nan, nan, nan],
[nan, nan, nan],
[nan, nan, nan],
[nan, nan, nan],
[nan, nan, nan],
[nan, nan, nan]], device='cuda:0', grad_fn=)

Thanks for the good work. Are you planning to open train/test code for RGB SOD? Also I found that there are no results on DUTS-TE, DUT-OMRON, ECSSD, HKU-IS, PASCAL-S which were reported on the paper. Will it be available sooner or later?

hello, where is your code???

How can I get the depth image processed by the DepthCorrectionNet?

perblom:ModuleNotFoundError: No module named 'ResNet'
When i first run test.py in win10,I ran into that problem.
Then I type "PIP install ResNet" and download it successfully, then I run it again and still encounter the same problem.
what can i do,:-)

Dear Jing:
I am very interested in your research, especially your new paper . Are you going to release its source code so that we can reproduce the results?

@JingZhang617 Hi! Thanks for your excellent works!
In the paper, you mentioned that you used hide-and-seek to generate five ground truths. In the code, the network learns from only one initial ground truth. Could you please give me a favor?

Besides, I generate results for test images using the provided models. I find the results for one image are the same. Could you tell me how to generate diverse outputs like Figure 8?

Looking forward to your answer. Thanks again!

i download the model_100_gen.pth and load it in the test.py
when i run the test.py,the error is :
size mismatch for sal_encoder.conv_depth1.conv.weight: copying a param with shape torch.Size([3, 9, 3, 3]) from checkpoint, the shape in current model is torch.Size([3, 14, 3, 3]).

The provided model in: https://drive.google.com/file/d/1nzGLnlmntTGbcaShfQvE6ouyfWJD-pIB/view?usp=sharing, is trained with k=3? Did you upload the wrong version?

Hi, could you please release the code for ABP? Thanks a lot!

The released code is using resnet50 as backbone. How about the provided saliency maps, vgg16 or resnet50? Can you also share the code of vgg version?

作者你好，测试代码这是怎么回事