backtime92

@zobeirraisi Sorry! I have released the code for IC15 just now.

@zobeirraisi Because it returns the confidence, you can change it as below: for index, (real_images, real_gh_label, real_gah_label, real_mask) in enumerate(real_data_loader): syn_images, syn_gh_label, syn_gah_label, syn_mask, __ = next(batch_syn)

@zobeirraisi Could you show me your whole error information

@zobeirraisi I think maybe is your image path error,please check your data path.

@ThisIsIsaac I had not noticed this bug, I will check it .Thanks for you comment.

@ThisIsIsaac @chenhaolin1989 sorry for the late reply, I have checkit ,but i did not find the bug. Can you share me more details.

Duplicate of #49

@ThisIsIsaac Q1: The output map down sample the input to 1/2 size Q2: Region and affinity score is between 0~1 Q3: Random crop the image to 768*768 I think you...

@ThisIsIsaac The 500 largest negative pixesl is not sensitive to the performance,you can choose 10000 or 40000 both ok. OHEM is just that the model choose positive an d negative...

@ThisIsIsaac The author freeze the model to train the linkRefiner (see the paper P11 and P12). When you train my reimplementation-model that you do not need freeze anything.