Rethinking Pre-Training and Augmentation for Zero-Shot Cross-City Object Detection

arXiv:2608.24154v1 Announce Type: cross Abstract: Real-world deployment of traffic surveillance systems is bottlenecked by geographic domain shift, in which models trained in one city underperform when applied to an unseen target city. Conventional domain adaptation relies on hyperparameter-sensitive architectures or direct profiling of target data. Both are fundamentally precluded in privacy-conscious ecosystems that require completely blind training and evaluation loops. In this setting, we explore the effects of pre-training and augmentation in addressing the domain shift problem. Specifically, we propose a new modular training pipeline for object detection structured around two core orthogonal pillars: (1) a multi-dataset pre-training strategy featuring a class-agnostic objectness distillation to decouple structural vehicle geometry from semantic taxonomies, and (2) a domain-resilient augmentation stream featuring a novel Grayworld transformation that forces global attention heads to strip volatile chromatic shortcuts in favor of robust shape priors. When evaluated with the real-time transformer-based detector RF-DETR, our framework bridges cross-city distribution gaps while using limited GPU memory (16GB). Our optimized variants, RF-DETR-HR and RF-DETR-Grayworld, deliver a substantial empirical gain of +24.29 over the baseline, achieving 1st place (47.53 mAP) on the AI City Challenge Track 6 leaderboard. Code and data are available at: href{https://github.com/SKKUAutoLab/aic26_cross_city}{SKKUAutoLab/aic26_cross_city}.

Related

• Hardware Utilization and Inference Performance of Edge Object Detection Under Fault Injection
• A Multi-Task Deep Learning Framework for Real-Time Intelligent Video Surveillance with Temporal Event Validation
• Closed-Form Linear-Probe Dataset Distillation for Pre-trained Vision Models

Source: arXiv cs.AI | 2026-08-26