lucidrains/pixel-level-contrastive-learning issues and pull requests

#20 - Could the pixel-level contrastive learning be adapted to 3D classification or segmentation task？

Issue - State: open - Opened by Lazzzycoding about 1 year ago

#19 - Apply PixPro to resnet50 with FPN

Issue - State: open - Opened by simonebonato about 1 year ago

#18 - Loss values are not reducing

Issue - State: open - Opened by NamburiSrinath about 3 years ago - 1 comment

#17 - Extending the model for Semantic Segmentation task

Issue - State: open - Opened by NamburiSrinath about 3 years ago - 3 comments

#16 - Could you report performance?

Issue - State: closed - Opened by jianlong-yuan almost 4 years ago

#15 - How to train PixelCL on multi-gpu?

Issue - State: open - Opened by ALEX13679173326 almost 4 years ago

#14 - positive pairs

Issue - State: open - Opened by jancylee almost 4 years ago - 5 comments

#13 - Same size models -> wildly different RAM usage

Issue - State: closed - Opened by makarr about 4 years ago - 2 comments

#12 - Added parameter for returning positive pixels pairs

Pull Request - State: closed - Opened by ahmed-bensaad about 4 years ago

#11 - Using pytorch-lightning to train PixelCL on multi-gpu

Issue - State: open - Opened by ahmed-bensaad about 4 years ago - 11 comments

#10 - Data Augumentation not same with BYOL?

Issue - State: closed - Opened by guanfuchen about 4 years ago - 3 comments

#9 - use keops in pixel propagation module

Pull Request - State: closed - Opened by lucidrains about 4 years ago

#8 - O(n^4) memory usage in PPM

Issue - State: closed - Opened by neonbjb about 4 years ago - 13 comments

#7 - about the pos-neg pair selection

Issue - State: closed - Opened by 1144181135 about 4 years ago - 1 comment

#6 - about distributed

Issue - State: closed - Opened by datuizhuang about 4 years ago - 1 comment

#5 - The distance_matrix and similarity_matrix may not match

Issue - State: closed - Opened by lygsbw about 4 years ago - 9 comments

#4 - Alpha - Combination of PixPro loss and PixCL

Issue - State: closed - Opened by WilhelmT about 4 years ago - 6 comments

#3 - Add the two loss functions instead of multiplying them

Pull Request - State: closed - Opened by cjrd about 4 years ago - 1 comment

#2 - Does using RandomResizedCrop/HorizontalFlip in the augmentaion pipeline break the pixel alignment?

Issue - State: closed - Opened by cjrd about 4 years ago - 4 comments

#1 - Help needed

Issue - State: closed - Opened by lucidrains about 4 years ago - 6 comments
Labels: help wanted

Ecosyste.ms: Issues

GitHub / lucidrains/pixel-level-contrastive-learning issues and pull requests