Contrastive Learning Network Based on Causal Attention for Fine-Grained Ship Classification in Remote Sensing Scenarios

Abstract

Fine-grained classification of ship targets is an important task in remote sensing, having numerous applications in military reconnaissance and sea surveillance. Due to the influence of various imaging factors, ship targets in remote sensing images have considerable inter-class similarity and intra-class difference, which brings significant challenges to fine-grained classification. In response, we developed a contrastive learning network based on causal attention (C2Net) to improve the model’s fine-grained identification ability from local details. The asynchronous feature learning mode of “decoupling + aggregation” is adopted to reduce the mutual influence between local features and improve the quality of local features. In the decoupling stage, the feature vectors of each part of the ship targets are de-correlated using a decoupling function to prevent feature adhesion. Considering the possibility of false associations between results and features, the decoupled part is designed based on the counterfactual causal attention network to enhance the model’s predictive logic. In the aggregation stage, the local attention weight learned in the decoupling stage is used to carry out feature fusion on the trunk feature weight. Then, the proposed feature re-association module is used to re-associate and integrate the target local information contained in the fusion feature to obtain the target feature vector. Finally, the aggregation function is used to complete the clustering process of the target feature vectors and fine-grained classification is realized. Using two large-scale datasets, the experimental results show that the proposed C2Net method had better fine-grained classification than other methods.

A Novel Multiscale Contrastive Learning Network for Fine-Grained Ocean Ship Classification

“…P 2 N et makes the features stay away from different classes and aggregate the features of the same class. Pan et al [1] developed a contrastive learning network (C2Net) for FGSC, which applies counterfactual causal reasoning to make decisions at the logical level and enhances attention to local details. Zhang et al [31] introduced a part assignment module and proposed a similarity learning by ranking (SIMLR) contrastive learning framework for FGSC.…”

“…Limited by the difficulty of data acquisition and the low quality of data [1], the study of ship classification initially focused on the target level and coarse granularity. These This work was supported by the National Natural Science Foundation of China (Grant No.…”

“…Therefore, some researchers focus on extracting the multiscale features for FGSC, which can improve the accuracy of the model with significant inter-class similarity and intra-class difference [15]- [22]. With the progress of contrastive learning, some methods were proposed to make the features of the same class closer by loss function [1], [23]. However, multi-scalebased methods either extract multi-scale features from the network, which can be influenced by complex backgrounds, or extract local features of the most attention regions of the feature map, which can be influenced by noise.…”