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

arXiv: https://arxiv.org/abs/2608.24154 · HTML: https://mlsanity.com/paper/2608.24154/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 27 of its 48 references that are in the map.
- First posted on arXiv 25 Aug 2026.

## Builds on (most cited first, 27 of 27)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks](https://mlsanity.com/paper/1506.01497/) (2015)
- [You Only Look Once: Unified, Real-Time Object Detection](https://mlsanity.com/paper/1506.02640/) (2015)
- [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://mlsanity.com/paper/2010.04159/) (2020)
- [Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detection](https://mlsanity.com/paper/2303.05499/) (2023)
- [Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond](https://mlsanity.com/paper/2308.12966/) (2023)
- [YOLO9000: Better, Faster, Stronger](https://mlsanity.com/paper/1612.08242/) (2016)
- [YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors](https://mlsanity.com/paper/2207.02696/) (2022)
- [Albumentations: fast and flexible image augmentations](https://mlsanity.com/paper/1809.06839/) (2018)
- [DETRs Beat YOLOs on Real-time Object Detection](https://mlsanity.com/paper/2304.08069/) (2023)
- [YOLO-World: Real-Time Open-Vocabulary Object Detection](https://mlsanity.com/paper/2401.17270/) (2024)
- [Domain Adaptive Faster R-CNN for Object Detection in the Wild](https://mlsanity.com/paper/1803.03243/) (2018)
- [MOT20: A benchmark for multi object tracking in crowded scenes](https://mlsanity.com/paper/2003.09003/) (2020)
- [DETRs with Collaborative Hybrid Assignments Training](https://mlsanity.com/paper/2211.12860/) (2022)
- [Detection and Tracking Meet Drones Challenge](https://mlsanity.com/paper/2001.06303/) (2020)
- [Domain-Specific Batch Normalization for Unsupervised Domain Adaptation](https://mlsanity.com/paper/1906.03950/) (2019)
- [FishEye8K: A Benchmark and Dataset for Fisheye Camera Object Detection](https://mlsanity.com/paper/2305.17449/) (2023)
- [The 8th AI City Challenge](https://mlsanity.com/paper/2404.09432/) (2024)
- [The 9th AI City Challenge](https://mlsanity.com/paper/2508.13564/) (2025)
- [Dynamic Retraining-Updating Mean Teacher for Source-Free Object Detection](https://mlsanity.com/paper/2407.16497/) (2024)
- [The 10th AI City Challenge](https://mlsanity.com/paper/2608.17044/) (2026)
- [TrafficCAM: A Versatile Dataset for Traffic Flow Segmentation](https://mlsanity.com/paper/2211.09620/) (2022)
- [Enhancing Thermal MOT: A Novel Box Association Method Leveraging Thermal Identity and Motion Similarity](https://mlsanity.com/paper/2411.12943/) (2024)
- [Vision Technologies with Applications in Traffic Surveillance Systems: A Holistic Survey](https://mlsanity.com/paper/2412.00348/) (2024)
- [SAGA: Semantic-Aware Gray color Augmentation for Visible-to-Thermal Domain Adaptation across Multi-View Drone and Ground-Based Vision Systems](https://mlsanity.com/paper/2504.15728/) (2025)
- [TSBOW -- Traffic Surveillance Benchmark for Occluded Vehicles Under Various Weather Conditions](https://mlsanity.com/paper/2602.05414/) (2026)

## Abstract (arXiv)

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: hrefhttps://github.com/SKKUAutoLab/aic26crosscitySKKUAutoLab/aic26crosscity.