FSDETR: Frequency-Spatial Feature Enhancement for Small Object Detection

TL;DR

FSDETR introduces a frequency-spatial feature enhancement framework that significantly improves small object detection by leveraging hierarchical attention, deformable intra-scale interactions, and frequency-spatial pyramid networks.

Contribution

The paper proposes a novel framework combining frequency filtering and spatial attention mechanisms to enhance small object detection performance.

Findings

Achieves 13.9% APS on VisDrone 2019 with 14.7M parameters.

Attains 48.95% AP50 tiny on TinyPerson dataset.

Outperforms existing methods on small-object benchmarks.

Abstract

Small object detection remains a significant challenge due to feature degradation from downsampling, mutual occlusion in dense clusters, and complex background interference. To address these issues, this paper proposes FSDETR, a frequency-spatial feature enhancement framework built upon the RT-DETR baseline. By establishing a collaborative modeling mechanism, the method effectively leverages complementary structural information. Specifically, a Spatial Hierarchical Attention Block (SHAB) captures both local details and global dependencies to strengthen semantic representation. Furthermore, to mitigate occlusion in dense scenes, the Deformable Attention-based Intra-scale Feature Interaction (DA-AIFI) focuses on informative regions via dynamic sampling. Finally, the Frequency-Spatial Feature Pyramid Network (FSFPN) integrates frequency filtering with spatial edge extraction via the Cross-domain Frequency-Spatial Block (CFSB) to preserve fine-grained details. Experimental results show that with only 14.7M parameters, FSDETR achieves 13.9% APS on VisDrone 2019 and 48.95% AP50 tiny on TinyPerson, showing strong performance on small-object benchmarks. The code and models are available at https://github.com/YT3DVision/FSDETR.