KPS Loss: Key Point Sensitive Loss for Long-tailed Visual Recognition

This is the source code for our TPAMI paper: Key Point Sensitive Loss for Long-tailed Visual Recognition.
This version is a demo of how to use KPS loss. The version that supports more datasets is in the works and is coming soon.

CIFAR10

$ python cifar_train_backbone.py --arch resnet32 /
                                 --dataset cifar10 --data_path './dataset/data_img'/
                                 --loss_type 'KPS' --train_rule 'GA'/
                                 --imb_factor 0.01/ 
                                 --batch_size 64 --learning_rate 0.1 

To do list:

• Support Cifar10/100-LT dataset
• Support imageNet-LT
• Support iNaturalist2018
• More loss functions
• Separate configuration files for easier execution
• Some other minor performance improvements

Citation

@article{Li2022Long,
  author    = {Mengke Li, Yiu{-}ming Cheung, Zhikai Hu},
  title     = {Key Point Sensitive Loss for Long-tailed Visual Recognition},
  journal   = {{IEEE} Transactions on Pattern Analysis and Machine Intelligence},
  volume    = {45},
  number    = {4},
  pages     = {4812 - 4825},
  publisher = {IEEE},
  year      = {2023},
  doi       = {10.1109/TPAMI.2022.3196044},
}

You May Find Our Additional Works of Interest

• [CVPR'22] Long-tailed visual recognition via Gaussian clouded logit adjustment [paper] [code]

• [CVPR'23] Long-Tailed Visual Recognition via Self-heterogeneous Integration with Knowledge Excavation [paper] [code]

• [AAAI'24] Feature Fusion from Head to Tail for Long-Tailed Visual Recognition [paper] [code]

• [NeurIPS'24] Improving Visual Prompt Tuning by Gaussian Neighborhood Minimization for Long-Tailed Visual Recognition [paper] [code]

• [TAI'24] Adjusting logit in Gaussian form for long-tailed visual recognition [paper] [code]

Other Resources of long-tailed visual recognition

• Awesome-LongTailed-Learning

• Awesome-of-Long-Tailed-Recognition

• Long-Tailed-Classification-Leaderboard

• BagofTricks-LT

Connection

If you have any questions, please send the email to Mengke Li at: csmkli@comp.hkbu.edu.hk.