Attribute Recognition from Adaptive Parts
Luwei Yang, Ligeng Zhu, Yichen Wei, Shuang Liang, Ping Tan · 2016
Previous part-based attribute recognition approaches perform part detection and attribute recognition in separate steps. The parts are not optimized for attribute recognition and therefore could be sub-optimal. We present an end-to-end deep learning approach to overcome the limitation. It generates object parts from key points and perform attribute recognition accordingly, allowing adaptive spatial transform [10] of the parts. Both key point estimation and attribute recognition are learnt jointly in a multi-task setting. Extensive experiments on two datasets verify the efficacy of the proposed end-to-end approach.