Semantic feature integration network for fine-grained visual classification

Fine-Grained Visual Classification (FGVC) is a challenging task due to subtle differences among subordinate categories. Many current FGVC approaches focus on identifying and locating discriminative regions, but neglect the presence of unnecessary features that impair the understanding of object structure. These unnecessary features, including 1) ambiguous parts resulting from the visual similarity in object appearances and 2) noninformative parts (e.g., background noise), can have a significant adverse impact on classification results. To address this limitation, we propose the Semantic Feature Integration network (SFI-Net) to eliminate unnecessary information and reconstructing the semantic relations among discriminative features. Our SFI-Net are carefully designed and can be trained end-to-end in a weakly-supervised way. Extensive experiments on four challenging fine-grained benchmarks demonstrate that our work achieves the state-of-thearts performance. Especially, the classification accuracy of our model on CUB-200-2011 and Stanford Dogs reaches 92.64% and 93.03%, respectively.