MGDFIS: Multi-scale Global-detail Feature Integration Strategy for Small Object Detection

2025/06/15 by Yuxiang Wang, Yu-Xiang Wang, Wang, Yuxiang +12
Computer Science · #Advanced Image and Video Retrieval Techniques #Artificial Intelligence (cs.AI) #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Machine Learning (cs.LG) #cs.AI #cs.CV #cs.LG

paper · pdf · doi:10.48550/arxiv.2506.12697

openalex publication_date 2025/06/15 · openalex created_date 2025/10/13 · openalex updated_date 2026/08/01 · arxiv created 2026/08/02 · arxiv updated 2026/08/04

Abstract

Small-object detection in Unmanned Aerial Vehicle (UAV) imagery requires preserving weak local evidence while using broader context to separate tiny foreground targets from cluttered backgrounds. Existing multi-scale fusion methods improve feature aggregation, but they often add computation or blur fine details during repeated cross-scale fusion. The central challenge is to balance low-SNR target preservation, clutter suppression, and efficient cross-scale context exchange. To address this challenge, we propose the Multi-scale Global-detail Feature Integration Strategy (MGDFIS), a neck-level feature-fusion strategy that couples global context exchange, local-detail recovery, and pixel-level foreground-background recalibration. MGDFIS integrates three coordinated modules: FusionLock-TSS Attention for stabilizing spectral-spatial responses, Global-detail Integration for combining long-range mixing with local detail capture, and Dynamic Pixel Attention for reweighting compact foreground regions. On the controlled VisDrone setting, YOLO26m + MGDFIS improves AP50:95 from 25.7 to 30.2 and AP50 from 37.2 to 44.2 over the YOLO26m baseline, with 96.1 GFLOPs. Additional dataset-specific evaluations report 38.9 AP50 and 21.9 AP50:95 on UAVDT and 97.4 AP50 on CARPK. The code is available at: https://github.com/JackBaixue/MGDFIS.

Citations

• Revisiting DETR for Small Object Detection via Noise-Resilient Query Optimization
• A Survey of 3D Reconstruction with Event Cameras
• Transformers without Normalization
• YOLOv12: Attention-Centric Real-Time Object Detectors
• UAV-DETR: Efficient End-to-End Object Detection for Unmanned Aerial Vehicle Imagery
• Towards End-to-End Neuromorphic Event-based 3D Object Reconstruction Without Physical Priors
• Token Statistics Transformer: Linear-Time Attention via Variational Rate Reduction
• HS-FPN: High Frequency and Spatial Perception FPN for Tiny Object Detection
• A Hybrid Transformer-Mamba Network for Single Image Deraining
• 5%>100%: Breaking Performance Shackles of Full Fine-Tuning on Visual Recognition Tasks
• Cross-Layer Feature Pyramid Transformer for Small Object Detection in Aerial Images
• A DeNoising FPN With Transformer R-CNN for Tiny Object Detection
• YOLOv10: Real-Time End-to-End Object Detection
• YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information
• A review of small object detection based on deep learning
• Adapter is All You Need for Tuning Visual Tasks
• Gold-YOLO: Efficient Object Detector via Gather-and-Distribute Mechanism
• DETRs Beat YOLOs on Real-time Object Detection
• DETRs Beat YOLOs on Real-time Object Detection
• Adaptive Sparse Convolutional Networks with Global Context Enhancement for Faster Object Detection on Drone Images
• Large Selective Kernel Network for Remote Sensing Object Detection
• BiFormer: Vision Transformer with Bi-Level Routing Attention
• Annotated Point Clouds and Images from a Spatial-Semantic Perception Pipeline: Robotized Deconstruction at the Reference Construction Site, Aachen
• YOLOv6: A Single-Stage Object Detection Framework for Industrial Applications
• Detecting tiny objects in aerial images: A normalized Wasserstein distance and a new benchmark
• DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection
• Vision Transformer with Deformable Attention
• Grounded Language-Image Pre-training
• QueryDet: Cascaded Sparse Query for Accelerating High-Resolution Small Object Detection
• Effective Fusion Factor in FPN for Tiny Object Detection
• Deformable DETR: Deformable Transformers for End-to-End Object Detection
• Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection
• Extended Feature Pyramid Network for Small Object Detection
• Detection and Tracking Meet Drones Challenge
• EfficientDet: Scalable and Efficient Object Detection
• EfficientDet: Scalable and Efficient Object Detection
• Clustered Object Detection in Aerial Images

Related

• MASF-YOLO: An Improved YOLOv11 Network for Small Object Detection on Drone View
• Efficient Feature Fusion for UAV Object Detection
• PT-DETR: Small Target Detection Based on Partially-Aware Detail Focus
• Small Object Detection: A Comprehensive Survey on Challenges, Techniques and Real-World Applications
• DGE-YOLO: Dual-Branch Gathering and Attention for Accurate UAV Object Detection