Multi-level Attention Network with Weather Suppression for All-Weather Action Detection in UAV Rescue Scenarios

Unmanned Aerial Vehicles (UAVs) possess significant advantages in terms of mobility and range compared to traditional surveillance cameras. Human action detection from UAV images has the potential to assist in various fields, including search and rescue operations. However, UAV images present challenges such as varying heights, angles, and the presence of small objects. Additionally, they can be affected by adverse illumination and weather conditions. In this paper, we propose a Multi-level Attention network with Weather Suppression for all-weather action detection in UAV rescue scenarios. The Weather Suppression module effectively mitigates the impact of illumination and weather, while the Multi-level Attention module enhances the model’s performance in detecting small objects. We conducted detection experiments under both normal and synthetic harsh conditions, and the results demonstrate that our model achieves state-of-the-art performance. Furthermore, a comparison of relevant metrics reveals that our model strikes a balance between size and complexity, making it suitable for deployment on UAV platforms. The conducted ablation experiments also highlight the significant contribution of our proposed modules.