Adapting Vehicle Detectors for Aerial Imagery to Unseen Domains with Weak Supervision

TL;DR

This paper introduces a generative AI-based multi-stage framework for adapting vehicle detectors to unseen aerial imagery domains, significantly improving detection performance across diverse environments.

Contribution

It presents a novel multi-modal knowledge transfer approach using fine-tuned latent diffusion models to bridge domain gaps in aerial vehicle detection.

Findings

Achieved 4-23% AP50 improvement over supervised source training.

Enhanced performance by 6-10% compared to weakly supervised adaptation.

Surpassed unsupervised domain adaptation and open-set detectors by over 50%.

Abstract

Detecting vehicles in aerial imagery is a critical task with applications in traffic monitoring, urban planning, and defense intelligence. Deep learning methods have provided state-of-the-art (SOTA) results for this application. However, a significant challenge arises when models trained on data from one geographic region fail to generalize effectively to other areas. Variability in factors such as environmental conditions, urban layouts, road networks, vehicle types, and image acquisition parameters (e.g., resolution, lighting, and angle) leads to domain shifts that degrade model performance. This paper proposes a novel method that uses generative AI to synthesize high-quality aerial images and their labels, improving detector training through data augmentation. Our key contribution is the development of a multi-stage, multi-modal knowledge transfer framework utilizing fine-tuned latent diffusion models (LDMs) to mitigate the distribution gap between the source and target environments. Extensive experiments across diverse aerial imagery domains show consistent performance improvements in AP50 over supervised learning on source domain data, weakly supervised adaptation methods, unsupervised domain adaptation methods, and open-set object detectors by 4-23%, 6-10%, 7-40%, and more than 50%, respectively. Furthermore, we introduce two newly annotated aerial datasets from New Zealand and Utah to support further research in this field. Project page is available at: https://humansensinglab.github.io/AGenDA