No such file or directory: './pretrain_model/resnet101-imagenet.pth' about self-correction-human-parsing HOT 5 CLOSED

@kpkroy

1. you have to download the pre-trained Imagenet model from the readme. Then place it inside the pretrain_model folder.
2. @PeikeLi correct me if I am wrong here, the simple_extractor is simply for visualization, from classes image-> color images, so you have to train the network got the model then use it to visualize, I believe it is written with the same purpose with the evaluate script
3. Yes, you are correct. However, I have found an issue in the training code that the schp.save_checkpoint has been changed to schp.save_schp_checkpoint. You need to fix it by yourself.

regards,

from self-correction-human-parsing.

@VyBui Thank you for replying and pointing out the issue.

In addition, simple_extractor.py aims for someone who only wants to get the parsing result or intermediate feature representations with the given pre-trained models.

from self-correction-human-parsing.

Thank you guys. Much appreciation.

from self-correction-human-parsing.

@kpkroy

1. you have to download the pre-trained Imagenet model from the readme. Then place it inside the pretrain_model folder.
2. @PeikeLi correct me if I am wrong here, the simple_extractor is simply for visualization, from classes image-> color images, so you have to train the network got the model then use it to visualize, I believe it is written with the same purpose with the evaluate script
3. Yes, you are correct. However, I have found an issue in the training code that the schp.save_checkpoint has been changed to schp.save_schp_checkpoint. You need to fix it by yourself.

regards,

Dear Sir

where to download resnet101-imagenet.pth you mentioned, I did not see any imagenet pretrain model from README.md

from self-correction-human-parsing.

Indeed I cannot find the pretrained weights either.
Is this something we can get from torchvision given it's a simple ResNet101?

Or is this the link?

from self-correction-human-parsing.