分類Global-View

Fine-Grained recognition require recognition of highly localized attributes of objects while being invariant to their pose and location in the image Part-based models construct representations by loca

>>阅读原文<<