DiffVAS: Diffusion-Guided Visual Active Search in Partially Observable Environments

2026-05-14

AI Summary

Existing visual active search methods rely on fully observable environments and support only single-object-category queries, making them ill-suited for real-world scenarios characterized by limited fields of view and the need to locate multiple target classes. To address these limitations, this work proposes DiffVAS, the first approach to integrate diffusion models into visual active search. DiffVAS employs a target-conditional policy to reconstruct a complete geographic map from sequential partial observations and leverages reinforcement learning for efficient unmanned aerial vehicle (UAV) path planning. By enabling unified, class-conditional search across multiple object categories, the method overcomes the constraints of full observability and single-target assumptions inherent in prior approaches. Extensive experiments demonstrate that DiffVAS significantly outperforms state-of-the-art methods across multiple datasets, achieving substantial gains in both accuracy and efficiency under partially observable conditions.

Abstract

Visual active search (VAS) has been introduced as a modeling framework that leverages visual cues to direct aerial (e.g., UAV-based) exploration and pinpoint areas of interest within extensive geospatial regions. Potential applications of VAS include detecting hotspots for rare wildlife poaching, aiding search-and-rescue missions, and uncovering illegal trafficking of weapons, among other uses. Previous VAS approaches assume that the entire search space is known upfront, which is often unrealistic due to constraints such as a restricted field of view and high acquisition costs, and they typically learn policies tailored to specific target objects, which limits their ability to search for multiple target categories simultaneously. In this work, we propose DiffVAS, a target-conditioned policy that searches for diverse objects simultaneously according to task requirements in partially observable environments, which advances the deployment of visual active search policies in real-world applications. DiffVAS leverages a diffusion model to reconstruct the entire geospatial area from sequentially observed partial glimpses, which enables a target-conditioned reinforcement learning-based planning module to effectively reason and guide subsequent search steps. Extensive experiments demonstrate that DiffVAS excels in searching diverse objects in partially observable environments, significantly surpassing state-of-the-art methods on several datasets.

Problem

• Visual Active Search
• Partially Observable Environments
• Multi-Target Search
• Geospatial Exploration
• Aerial Surveillance

Innovation

• diffusion model
• visual active search
• partially observable environment
• target-conditioned policy
• reinforcement learning