UAV-C

UAV-C是一个大规模的无人机跟踪基准数据集，由北德克萨斯大学和普渡大学联合创建。该数据集基于两个流行的无人机数据集UAV123和DTB70，引入了18种常见的破坏类型，涵盖了天气、传感器、模糊和复合破坏等四个类别。UAV-C包含超过9843个视频序列，总计超过310万帧，旨在评估无人机跟踪器在面对各种破坏时的鲁棒性。数据集的创建过程涉及对原始视频帧应用特定的算法来模拟不同的破坏情况。UAV-C的应用领域主要集中在提高无人机在复杂环境下的跟踪性能，特别是在监控和机器人技术中。UAV-C is a large-scale benchmark dataset for unmanned aerial vehicle (UAV) tracking, jointly developed by the University of North Texas and Purdue University. Built upon two popular UAV tracking datasets, UAV123 and DTB70, this dataset introduces 18 common corruption types, covering four categories: weather-related, sensor-related, motion blur, and compound corruptions. UAV-C contains over 9843 video sequences with a total of more than 3.1 million frames, and it is designed to evaluate the robustness of UAV trackers under various corruptions. The construction of this dataset involves applying specific algorithms to the original video frames to simulate different corruption scenarios. The primary application scenarios of UAV-C focus on improving the tracking performance of UAVs in complex environments, especially in the fields of surveillance and robotics.

搜集汇总

背景与挑战

背景概述

UAV-C是一个用于评估无人机跟踪方法鲁棒性的基准数据集，它基于DTB70和UAV123_10fps数据集，通过应用18种常见腐蚀类型（每种有3个严重级别）来模拟图像质量受损情况。该数据集旨在测试跟踪算法在真实世界腐蚀条件下的性能，为研究提供标准化评估工具。