MMA-UAV：A multimodal low altitude anti-UAV dataset based on infrared/visible light images, active radar, and passive radio frequency signals.

Citation Author(s):: Chang Bo Hou

Xian Ye Ben

Bin Wang (School of Information and Communication Engineering, Harbin Engineering University,Harbin,150001,China.)

Xi Lai Chen (School of Information and Communication Engineering, Harbin Engineering University,Harbin,150001,China.)

Si Cheng Liu (School of Information and Communication Engineering, Harbin Engineering University,Harbin,150001,China.)

Rui Qi Wang (School of Information and Communication Engineering, Harbin Engineering University,Harbin,150001,China.)

Yao Hui Tang (School of Information and Communication Engineering, Harbin Engineering University,Harbin,150001,China.)

Zhi Chao Zhou (School of Information and Communication Engineering, Harbin Engineering University,Harbin,150001,China.)
Submitted by:: Bin Wang
Last updated:: Fri, 10/24/2025 - 02:19
DOI:: 10.21227/wn7q-rg08
Data Format:: ZIP

429 views

Categories:

Artificial Intelligence

Keywords:

Anti-UAV

CITE

Abstract

This is the preliminary version (v0.1) of the MMA-UAV dataset (Multimodal Anti-UAV Dataset). The full version (v1.0) will be released later along with the corresponding publication.

In order to effectively overcome the inherent performance bottleneck and detection blind spots of a single sensor in unmanned aerial vehicle detection, recognition, and tracking, we introduce passive radio frequency mode for the first time on the basis of visual mode and radar mode. Through enzyme action intelligent optimization algorithm, we overcome the difficulty of large computation in radio frequency mode, effectively compensate for the shortcomings of visual and radar modes in long-distance small target detection, and construct the first multimodal low altitude anti unmanned aerial vehicle dataset integrating "infrared/visible light images+active radar+passive radio frequency signals".