Oriented Feature Alignment for Fine-grained Object Recognition in High-Resolution Satellite Imagery
This work addresses the problem of distinguishing fine-grained objects in complex remote sensing scenes for applications like satellite image analysis, but it appears incremental as it builds on existing oriented object detection methods.
The paper tackles fine-grained object recognition in high-resolution satellite imagery by proposing an oriented feature alignment network (OFA-Net), which achieved a mAP of 46.51% in the GaoFen competition and 43.73% in the ISPRS benchmark.
Oriented object detection in remote sensing images has made great progress in recent years. However, most of the current methods only focus on detecting targets, and cannot distinguish fine-grained objects well in complex scenes. In this technical report, we analyzed the key issues of fine-grained object recognition, and use an oriented feature alignment network (OFA-Net) to achieve high-performance fine-grained oriented object recognition in optical remote sensing images. OFA-Net achieves accurate object localization through a rotated bounding boxes refinement module. On this basis, the boundary-constrained rotation feature alignment module is applied to achieve local feature extraction, which is beneficial to fine-grained object classification. The single model of our method achieved mAP of 46.51\% in the GaoFen competition and won 3rd place in the ISPRS benchmark with the mAP of 43.73\%.