Leaf cultivar identification via prototype-enhanced learning

Yiyi Zhang, Zhiwen Ying, Ying Zheng, Cuiling Wu, Nannan Li, Fangfang Wang, Jun Wang, Xianzhong Feng, Xiaogang Xu

Abstract: Highlights•A novel method named Prototype-enhanced Learning is proposed for leaf cultivar identification.•We design a new prototype update module to learn inter-class relations for soft targets generation.•Extensive experiments on 7 benchmarks adequately illustrate the superiority of PEL for UFGVC.