LibFewShot is a comprehensive library for few-shot learning (FSL), especially for few-shot image classification. It integrates multiple classic FSL methods into a unified framework, including four fine-tuning based methods, six meta-learning based methods, and eight metric-learning based methods. This library is friendly for beginners in few-shot learning with very concise code and clear structure.

LibFewShot: A Comprehensive Library for Few-shot Learning. Wenbin Li, Chuanqi Dong, Pinzhuo Tian, Tiexin Qin, Xuesong Yang, Ziyi Wang, Jing Huo, Yinghuan Shi, Lei Wang, Yang Gao, Jiebo Luo. In arXiv 2021.


Special acknowledgment is first given to LegendDong, who built the foundation of this library and completed the most of algorithms. The following excellent contributors also participated in the development of this library throughout the process: WenbinLee, yangcedrus, wZuck, WonderSeven, Pinzhuo Tian, onlyyao, and cjy97.