Year: 2024
Type: conference-paper
Abstract: In this paper, we propose a small UAV detection system based on the DetUAV-Mamba model, which integrates a state-space model into the YOLO framework to improve the identification of small targets. The DetUAV-Mamba model is designed for high-resolution images and enables fast and accurate detection of micro-UAVs, which is crucial for modern security applications. By combining Mamba with an attention mechanism, the model significantly improves convergence speed and detection accuracy. Experimental results show that the DetUAV-Mamba model outperforms other models on both the UAV tracking dataset and the UAV410 dataset, demonstrating its potential for real-world aerial surveillance.
Authors: Mengqi Zhang, Huang Yan, Liwang Zhang, Zhuoyu Lin, Shaowu Yang
Institution: National University of Defense Technology
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4409832947