search

wsxtyrdd / NTSCC_JSAC22

71 stars 15 forks source link

"When I run python main.py --phase train, the following error occurs." #7

Closed jiangnan415 closed 3 months ago

jiangnan415 commented 4 months ago

Traceback (most recent call last): File "/home/jn/SC_work/NTSCC_JSAC22-master/main.py", line 175, in main(sys.argv[1:]) File "/home/jn/SC_work/NTSCC_JSAC22-master/main.py", line 163, in main loss = test(net, test_loader, logger) File "/home/jn/SC_work/NTSCC_JSAC22-master/main.py", line 25, in test mse_loss_ntc, bpp_y, bpp_z, mse_loss_ntscc, cbr_y, x_hat_ntc, x_hat_ntscc = net(input_image) File "/home/jn/miniconda3/envs/ntscc/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(input, kwargs) File "/home/jn/SC_work/NTSCC_JSAC22-master/net/NTSCC_Hyperior.py", line 119, in forward self.forward_NTC(input_image, require_probs=True) File "/home/jn/SC_work/NTSCC_JSAC22-master/net/NTSCC_Hyperior.py", line 152, in forward_NTC return super(NTSCC_Hyperprior, self).forward(input_image, kwargs) File "/home/jn/SC_work/NTSCC_JSAC22-master/net/NTSCC_Hyperior.py", line 53, in forward y = self.ga(input_image) File "/home/jn/miniconda3/envs/ntscc/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(input, kwargs) File "/home/jn/SC_work/NTSCC_JSAC22-master/layer/analysis_transform.py", line 98, in forward x = layer(x) File "/home/jn/miniconda3/envs/ntscc/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, *kwargs) File "/home/jn/SC_work/NTSCC_JSAC22-master/layer/analysis_transform.py", line 35, in forward x = self.downsample(x) File "/home/jn/miniconda3/envs/ntscc/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(input, kwargs) File "/home/jn/SC_work/NTSCC_JSAC22-master/layer/layers.py", line 363, in forward assert H % 2 == 0 and W % 2 == 0, f"x size ({H}{W}) are not even." AssertionError: x size (6811024) are not even.

image
wsxtyrdd commented 4 months ago

Please make sure the image size in your training dataset is the multipy of 64. If you need to train NTSCC in tiny images, e.g., CIFAR10, please change the the downsampling factor of analysis/synthesis transform.