ARD100
收藏资源简介:
ARD100数据集由西湖大学人工智能系创建,包含100个视频序列,总计202,467帧,采用DJI Mavic2或DJI M300相机在低中度海拔拍摄。该数据集涵盖了复杂背景、低光与强光条件、突然的相机移动、快速移动的无人机以及极小的无人机等多种现实世界挑战。数据集中的目标平均尺寸仅为帧大小的0.01%,是目前已有空对空无人机检测数据集中目标尺寸最小的一个。
The ARD100 dataset was created by the Department of Artificial Intelligence at Westlake University. It includes 100 video sequences with a total of 202,467 frames, and was captured using DJI Mavic 2 or DJI M300 cameras at low to moderate altitudes. This dataset covers a variety of real-world challenges, such as complex backgrounds, low-light and high-light conditions, sudden camera movements, fast-moving unmanned aerial vehicles (UAVs), and extremely small UAVs. The average size of the targets in the dataset is only 0.01% of the frame size, making it the dataset with the smallest target dimensions among currently available air-to-air UAV detection datasets.

- 1YOLOMG: Vision-based Drone-to-Drone Detection with Appearance and Pixel-Level Motion Fusion西湖大学人工智能系 · 2025年



