oneTaken

same behavior on M1 Pro

@GuoShi28 Thanks for your reply. So, it's a choice to choose which way. Can I ask why you change the implement function? Is the original ways before is not good...

Just the case in the `README` file. the 13-th case. https://github.com/Ryanshuai/BM3D_py#comparison-of-cpp-version-and-python-version Maybe some tricks what you explain above is reasonable.

@lyp0413 Do you check the python & tensorflow environment and the input image shape?

@gaosanyuan This method is not helpful for RGB only inputs. You can get from the results of the paper that if change the raw data into RGB data, the results...

The crop operation does not work during the model. The crop operation work before the training. Every training iteration, the origin point randomly selected and the crop image is generated...

I mean that the image saving code: ``` scipy.misc.toimage(temp * 255, high=255, low=0, cmin=0, cmax=255).save( result_dir + '%04d/%05d_00_train_%d.jpg' % (epoch, train_id, ratio)) ``` It's really slow and the `cv2.imwrite` is...

When evaluating, this certain one image can be removed easily. But the **PSNR** will increase, right? With your pretrained model, the results before and after removing the 10034 pairs are...

@CQFIO @cchen156 would you provide the solved dataset if you solved this problem? By the way, what's your following schedule? Is there some following paper published, considering the open problems...

Thanks for your check. Besides, did you check the train & val datasets?