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.

Remote sensing ship detection is a hotspot in computer vision, which is vital in both military and civilian fields. Nevertheless, the arbitrary orientation and dense arrangement of ship targets impose significant challenges in high-precision detection. Although research on this problem has progressed, high-precision detection is still limited by angular prediction accuracy. To tackle the issue, we start with angle prediction and propose a Dynamic Adjusting Labels (DAL) strategy based on Binary Coded Label (BCL). DAL strategy dynamically adjusts the ground-truth coded labels in the training process to guide angle coding for tendency learning. This strengthens the coupling between the angle coding bits and improves the performance of small granularity intervals. Due to the angle interval granularity difference, the learning difficulty of the coding layers and the convergence speed vary significantly. Aiming at this problem, we add a gradient truncation mechanism to each coding bit loss. The mechanism can effectively balance the coding layers’ learning strength and enhance the model’s training emphasis on coding bits corresponding to small granularity intervals, avoiding the effect of coding bits learning imbalance on angle prediction. Extensive experiments based on three public datasets demonstrate our method’s superiority in high-precision detection and state-of-the-art performance. Our code is available at https://github.com/lirunsheng2008/SDALS.