search

Owen718 / ECCV22-Perceiving-and-Modeling-Density-for-Image-Dehazing

ECCV'22 Oral | Perceiving and Modeling Density for Single Image Dehazing.
MIT License
55 stars 6 forks source link

test code #8

Closed zhenyuan1234 closed 1 year ago

zhenyuan1234 commented 1 year ago

Hello, the test code is not complete, can you release the complete code?

Owen718 commented 1 year ago

Hi The test code is complete. Could you please describe the problem you meet ?

zhenyuan1234 commented 1 year ago

I had an error like this when I was testing, thanks: File "G:\ECCV22-Perceiving-and-Modeling-Density-for-Image-Dehazing-main\Inference_Code\test.py", line 42, in Net=torch.jit.load(opt.pk_path,map_location=torch.device('cuda:0')).eval().cuda() File "G:\anaconda3\envs\pytorch\lib\site-packages\torch\jit_serialization.py", line 161, in load cpp_module = torch._C.import_ir_module(cu, str(f), map_location, _extra_files) RuntimeError: Unknown type name 'NoneType': Serialized File "code/torch/model/TSANet16.py", line 156 def forward(self: torch.model.TSANet16.MixScale, argument_1: Tensor, argument_2: Tensor) -> NoneType:


_53 = self.factor
_54 = (self.mixblock).forward(_53, )
zhenyuan1234 commented 1 year ago

Can you tell me where the .py file for the model

Owen718 commented 1 year ago

The TJM-based inference don't need any python file of model. Please refer to the readme.md .

Owen718 commented 1 year ago

Please confirm that the tjm model has been loaded successfully and image path is correct.