TinyDef-DETR: A Transformer-Based Framework for Defect Detection in Transmission Lines from UAV Imagery

TL;DR

TinyDef-DETR tackles the problem of detecting tiny transmission-line defects in UAV imagery by integrating four key components into a DETR-based framework: an Edge-Enhanced ResNet backbone, a stride-free Space-to-Depth downsampling module, a Cross-Stage Dual-Domain Multi-Scale Attention Module, and a Focaler-Wise-SIoU loss. The approach strengthens boundary-sensitive representations, preserves pixel-level details during downsampling, fuses global and local cues through frequency-spatial attention, and stabilizes small-object localization with a dynamic, difficulty-aware regression objective. On the CSG-ADCD dataset and real-world UAV data, TinyDef-DETR achieves superior accuracy and recall with modest computational overhead, and ablation studies confirm that each module contributes complementary benefits. Generalization tests on VisDrone-DET 2019 demonstrate cross-domain robustness for small-object detection in aerial imagery, indicating practical utility for UAV-based inspection and potential extension to other small-object detection tasks.

Abstract

Automated defect detection from UAV imagery of transmission lines is a challenging task due to the small size, ambiguity, and complex backgrounds of defects. This paper proposes TinyDef-DETR, a DETR-based framework designed to achieve accurate and efficient detection of transmission line defects from UAV-acquired images. The model integrates four major components: an edge-enhanced ResNet backbone to strengthen boundary-sensitive representations, a stride-free space-to-depth module to enable detail-preserving downsampling, a cross-stage dual-domain multi-scale attention mechanism to jointly model global context and local cues, and a Focaler-Wise-SIoU regression loss to improve the localization of small and difficult objects. Together, these designs effectively mitigate the limitations of conventional detectors. Extensive experiments on both public and real-world datasets demonstrate that TinyDef-DETR achieves superior detection performance and strong generalization capability, while maintaining modest computational overhead. The accuracy and efficiency of TinyDef-DETR make it a suitable method for UAV-based transmission line defect detection, particularly in scenarios involving small and ambiguous objects.

Figures (10)

• Figure S1: Overall framework of TinyDef-DETR
• Figure S2: The architecture of EEBlock (Edge Enhanced Block)
• Figure S3: Overall framework of EE-ResNet
• Figure S4: Illustration of SPDConv (Space-to-Depth Convolution)
• Figure S5: The architecture of the proposed Cross-Stage Dual-Domain Multi-Scale Attention Module (CSDMAM).