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

2025-07-28

AI Summary

To address the domain shift problem—characterized by weak cross-region generalization in aerial vehicle detection due to environmental, urban layout, and imaging condition variations—this paper proposes a multi-stage, multimodal knowledge transfer framework. It leverages a fine-tuned latent diffusion model (LDM) to synthesize high-fidelity remote sensing imagery and corresponding pseudo-labels, integrated with generative data augmentation and weakly supervised cross-domain adaptation for end-to-end detection optimization without dense annotations. The core contribution is a generative-AI-driven, weakly supervised domain alignment mechanism that significantly reduces the distribution discrepancy between source and target domains. Extensive experiments on multiple public benchmarks demonstrate consistent improvements, with AP₅₀ gains of 4–23% over state-of-the-art supervised, weakly supervised, unsupervised domain adaptation, and open-set detection methods. Additionally, two newly annotated aerial image datasets—covering New Zealand and Utah—are publicly released to support future research.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Remote Sensing / Geospatial AI
• Application Domains: Mobility, Driving & Flight

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

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

Problem

• Improving vehicle detection in unseen aerial domains
• Addressing domain shifts in geographic regions
• Enhancing detector training with generative AI

Innovation

• Generative AI synthesizes aerial images and labels
• Multi-stage multi-modal knowledge transfer framework
• Fine-tuned latent diffusion models reduce domain gaps