mlsanityData to 14 Oct 2026

FSDETR: Frequency-Spatial Feature Enhancement for Small Object Detection

arXiv 2604.14884 ↗First posted 16 Apr 2026cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

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.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 22 of its 30 references that are in the map. First posted on arXiv 16 Apr 2026.

Builds on

22 of 30 references.

References · 22

  1. 2016Feature Pyramid Networks for Object Detectioncited by 5,036
  2. 2020Deformable DETR: Deformable Transformers for End-to-End Object Detectioncited by 3,422
  3. 2021Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutionscited by 1,962
  4. 2022DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detectioncited by 1,319
  5. 2017Cascade R-CNN: Delving into High Quality Object Detectioncited by 1,276
  6. 2018Path Aggregation Network for Instance Segmentationcited by 798
  7. 2023DETRs Beat YOLOs on Real-time Object Detectioncited by 533
  8. 2024YOLOv10: Real-Time End-to-End Object Detectioncited by 515
  9. 2019CSPNet: A New Backbone that can Enhance Learning Capability of CNNcited by 381
  10. 2025YOLOv12: Attention-Centric Real-Time Object Detectorscited by 305
  11. 2021Global Filter Networks for Image Classificationcited by 260
  12. 2020VarifocalNet: An IoU-aware Dense Object Detectorcited by 231
10 more
  1. 2023Run, Don't Walk: Chasing Higher FLOPS for Faster Neural Networkscited by 130
  2. 2024RT-DETRv2: Improved Baseline with Bag-of-Freebies for Real-Time Detection Transformercited by 91
  3. 2019FreeAnchor: Learning to Match Anchors for Visual Object Detectioncited by 90
  4. 2021TPH-YOLOv5: Improved YOLOv5 Based on Transformer Prediction Head for Object Detection on Drone-captured Scenarioscited by 83
  5. 2024D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinementcited by 74
  6. 2019Scale Match for Tiny Person Detectioncited by 61
  7. 2021Focal and Efficient IOU Loss for Accurate Bounding Box Regressioncited by 57
  8. 2022RFLA: Gaussian Receptive Field based Label Assignment for Tiny Object Detectioncited by 51
  9. 2022No More Strided Convolutions or Pooling: A New CNN Building Block for Low-Resolution Images and Small Objectscited by 27
  10. 2025Rethinking Features-Fused-Pyramid-Neck for Object Detectioncited by 5

Built on by

Not cited yet.