Real-Time Source-Free Object Detection

2026-06-30

AI Summary

This work addresses the challenge of simultaneously achieving real-time performance, model lightweighting, and high accuracy in source-free domain adaptive object detection. Building upon the YOLOv10 dual-head architecture, the authors propose an efficient domain adaptation method that requires no modification to the inference structure. The approach introduces two key innovations: a DHF pseudo-label fusion strategy that enhances supervision quality by selectively leveraging one-to-one (O2O) and one-to-many (O2M) predictions, and a MARD multi-scale feature diversification loss that improves feature discriminability. Evaluated across multiple domain shift benchmarks, the method achieves mAP gains of 1.4–3.5%, a 1.3× increase in throughput, and nearly 50% reduction in parameter count, significantly outperforming existing source-free approaches.

Abstract

Real-world detectors for autonomous driving, surveillance, and robotics must handle domain-shifts under strict latency and memory constraints, yet existing source-free object detection (SFOD) methods rely on heavyweight architectures that prioritize accuracy alone. We show this trade-off is unnecessary: building on YOLOv10, an NMS-free dual-head detector, we achieve state-of-the-art adaptation accuracy while being faster and more compact. We observe that directly applying vanilla mean-teacher self-training to dual-head detectors leads to suboptimal adaptation performance due to two key factors. First, simple pseudo-label generation strategies, such as using a single head or directly combining high-confidence predictions from both heads, yield suboptimal supervision under domain-shift. We propose DHF (Dual-Head Pseudo-Label Fusion) which selectively admits one-to-one (O2O) and one-to-many (O2M) head predictions, preserving precision and recovering missed objects. Second, we observe domain-shift collapses multi-scale feature discriminability. We propose the use of our MARD (Multi-scale Adaptive Representation Diversification) loss which mitigates this by enforcing detection-aware variance and covariance constraints on multi-scale feature maps. Both modules are training-time only, leaving inference unchanged. Across domain-shift benchmarks, our method, RT-SFOD yields 1.4 to 3.5\% mAP gains, 1.3$\times$ higher throughput, with $\sim$2$\times$ fewer parameters than prior state-of-the-art SFOD methods, thus advancing the Pareto frontier of the speed-accuracy-model size trade-off. We report main results with YOLOv10, and demonstrate generalizability with additional YOLO- and DETR-based dual-head detectors. Code is available here: https://github.com/Sairam13001/RT-SFOD/

Problem

• Source-Free Object Detection
• Domain Shift
• Real-Time Inference
• Model Efficiency
• Autonomous Driving

Innovation

• Source-Free Object Detection
• Dual-Head Pseudo-Label Fusion
• Multi-scale Adaptive Representation Diversification
• Domain Adaptation
• Real-Time Detection