Conversation
|
Hi, Thanks for mentioning the 'labelcol' issue. The datasets used in this work have a labelcol folder. So, I think it is better to change the readme. Regarding resizing, for Monuseg dataset, we conduct experiments at 512x512 image scale. So, we pass the image size also as an argument. So, I guess it would be better if we provide an argument for resize and then convert all the images to imgsize if resize argument is true, which would make the code more generic rather than resizing everything to 128x128. Would be glad to merge a pull request with that ! |
|
Hi,The size of the image in the GlaS dataset is (522,775). Does it need to be changed to (128,128) or make some other adjustments |
|
hello, I want to ask you a question. I have been reporting this error during training. my loss has been 0 ,What's the matter about the error? Can you provide a processed data set? my email is 2536497516@qq.com 。 Can I contact you through this email? |
|
Hi, author! I would like to know how to conduct the experiment with a 512x512 image ratio in the Monuseg dataset. How did you adjust the image size? I urgently need to solve this problem in my experiment. Looking forward to your reply. Thanks! |
'labelcol' is not the folder name as in your readme.md, thus, code break. I changed it to 'label'. Also, the default size you have kept is (128, 128) but if images with the user are not uniform then it is an issue, I have rectified that by resizing the images to (128, 128).