Related papers: Oriented Feature Alignment for Fine-grained Object Recognition in High-Resolution Satellite Imagery

Oriented Feature Alignment for Fine-grained Object Recognition in High-Resolution Satellite Imagery

URL: http://arxiv.org/abs/2110.06628v1
Date: Wed, 13 Oct 2021 10:48:11 GMT
Title: Oriented Feature Alignment for Fine-grained Object Recognition in High-Resolution Satellite Imagery
Authors: Qi Ming, Junjie Song, Zhiqiang Zhou
Abstract summary: We analyze the key issues of fine-grained object recognition, and use an oriented feature alignment network (OFA-Net) to achieve high-performance object recognition. OFA-Net achieves accurate object localization through a rotated bounding boxes refinement module. 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%.
Score: 1.0635248457021498
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Abstract: 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\%.