Learning Class Unique Features in Fine-Grained Visual Classification
This work provides an incremental improvement for researchers and practitioners working on Fine-Grained Visual Classification, specifically addressing the issue of inter-class invariant features.
This paper addresses the challenge of distinguishing highly similar categories in Fine-Grained Visual Classification (FGVC) by proposing a novel regularization method. The authors formulate a minimax loss based on a game-theoretic framework, complemented by a Feature Redundancy Loss, to enforce the uniqueness of features to each category. Their method improves baseline model performance without additional computation and achieves results comparable to state-of-the-art models on several benchmarks.
A major challenge in Fine-Grained Visual Classification (FGVC) is distinguishing various categories with high inter-class similarity by learning the feature that differentiate the details. Conventional cross entropy trained Convolutional Neural Network (CNN) fails this challenge as it may suffer from producing inter-class invariant features in FGVC. In this work, we innovatively propose to regularize the training of CNN by enforcing the uniqueness of the features to each category from an information theoretic perspective. To achieve this goal, we formulate a minimax loss based on a game theoretic framework, where a Nash equilibria is proved to be consistent with this regularization objective. Besides, to prevent from a feasible solution of minimax loss that may produce redundant features, we present a Feature Redundancy Loss (FRL) based on normalized inner product between each selected feature map pair to complement the proposed minimax loss. Superior experimental results on several influential benchmarks along with visualization show that our method gives full play to the performance of the baseline model without additional computation and achieves comparable results with state-of-the-art models.