[CVPR2026] HAD: Heterogeneity-Aware Distillation for Lifelong Heterogeneous Learning

This repository contains the code and supplementary material for our CVPR 2026 paper.

Requirements

• PyTorch
• torchvision
• numpy
• opencv-python
• pillow
• scipy
• scikit-image
• six
• tqdm
• matplotlib

Datasets

Currently, we provide the implementation for the NYUv2 dataset. The code for the remaining datasets is being organized and will be released as soon as possible.

Please arrange the dataset as follows:

/data/dataset/nyuv2/
├── train/
│   ├── image/
│   ├── label/
│   ├── depth/
│   └── normal/
└── val/
    ├── image/
    ├── label/
    ├── depth/
    └── normal/

The dataset path is set in datasetting/continual_dataset.py.

Running scripts

$ python main.py

Acknowledgment

Our project references the codes in the following repos.

• MTSAM
• MAMMOTH
• SGP
• SPG

Citation

If you find our work useful in your research, please consider citing our paper.

@inproceedings{had,
  title={HAD: Heterogeneity-Aware Distillation for Lifelong Heterogeneous Learning},
  author={Zhang, Xuerui and Wang, Xuehao and Zhuang, Zhan and Zhao, Linglan and Li, Ziyue and Zhang, Xinmin and Song, Zhihuan and Zhang, Yu},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={10863--10873},
  year={2026}
}