Bidirectional Matching Prototypical Network for Few-Shot Image Classification

Abstract

Few-shot image classification (FSIC) is the task of generalizing a model to unknown categories by learning from a small number of labeled samples of some given categories. Recently, metric-based approaches have received lots of attention for their simplicity and effectiveness, but they often only use support set to generate inaccurate prototypes matching query set, ignoring the rich information contained in queries and the reversibility of the matching relationship between the two. In this letter, we propose a new simple and effective metric-based method called Bidirectional Matching Prototypical Network (BMPN), which has three innovations:1)It has an additional reverse matching process. This process uses queries to generate more accurate prototypes to improve the model’s performance while also forcing the model to learn features far from the decision boundary to enhance generalization capabilities; 2)It has a lightweight coordinate attention feature extractor (CAFE). This module not only captures long-term dependence along one spatial direction but also saving the accurate position information of the other spatial direction, helping the model to locate the region of interest more accurately; 3)It has a joint loss function, including forward matching loss and reverse matching loss, and a progressive weighting strategy is used in the training process to balance the importance of the two. Our model is trained end-to-end, and the experimental results show that we have reached the most advanced performance on the two benchmark datasets.