Fine-grained image recognition is very challenging due to the difficulty of capturing both semantic global features and discriminative local features. Meanwhile, these two features are not easy to be integrated, which are even conflicting when used simultaneously. In this paper, a retrieval-based coarse-to-fine framework is proposed, where we re-rank the TopN classification results by using the local region enhanced embedding features to improve the Top1 accuracy (based on the observation that the correct category usually resides in TopN results). To obtain the discriminative regions for distinguishing the fine-grained images, we introduce a weakly-supervised method to train a box generating branch with only image-level labels. In addition, to learn more effective semantic global features, we design a multi-level loss over an automatically constructed hierarchical category structure. Experimental results show that our method achieves state-of-the-art performance on three benchmarks: CUB-200-2011, Stanford Cars, and FGVC Aircraft. Also, visualizations and analysis are provided for better understanding.

翻译：精细的图像识别非常具有挑战性,因为很难捕捉到语义全球特征和歧视性的地方特征。 同时,这两个特征并不容易整合,在同时使用时甚至相互冲突。在本文中,提出了基于检索的粗粗到软框架,我们通过使用本地区域增强的嵌入功能来重新排序“顶部”分类结果,以提高“顶部”的准确性(基于正确类别通常位于“顶部”结果的观察)。为了获得区分精细图像的受歧视区域,我们采用了一种薄弱的监管方法来培训一个只有图像级标签的盒子生成分支。此外,为了学习更有效的语义全球特征,我们设计了一个针对自动构建的等级分类结构的多层次损失。实验结果表明,我们的方法在三个基准(CUB-200-2011年、斯坦福汽车和FGVC航空器)上达到了“最新”的性能。此外,为更好地理解提供了可视化和分析。