DAVAR-Lab-ML is a repository of machine learning algorithms from Davar Lab. The researches of this repository have included/will include knowledge distillation, semi-supervised learning, active learning, incremental learning, learning with noisy labels and so on.
- Extensibility: The algorithms are well suited for MM-series repository, e.g., MMDetection, MMClassfication...
- Plug-and-Play: The modules or operations are easily incorporated with different deep learning tasks, e.g., image classification, object detection, image segmentations...
- Easy-to-Use: The algorithms in this repository are well tested, and they can be easily run in your machines.
- The knowledge distillation framework Disitilling Object Detectors with Gloabl Knowledge. ECCV, 2022.
The source codes of two accepted papers on CVPR2023.
- 2022/07/30, release source code of the knowledge distillation algorithm Disitilling Object Detectors with Gloabl Knowledge.