DTB70, UAVDT, UAV123
收藏资源简介:
本研究使用了三个无人机跟踪数据集:DTB70、UAVDT和UAV123,这些数据集均记录于30 FPS,用于评估视觉跟踪算法在真实世界场景中的性能。DTB70包含70个视频序列,主要用于评估算法在高速移动目标上的跟踪能力。UAVDT和UAV123则提供了更广泛的场景和目标类型,以增强评估的全面性。这些数据集的创建旨在解决无人机视觉跟踪中的实时性和准确性问题,特别是在处理延迟和计算资源有限的情况下。
This study utilizes three drone tracking datasets: DTB70, UAVDT, and UAV123, all recorded at 30 FPS, to evaluate the performance of visual tracking algorithms in real-world scenarios. DTB70 contains 70 video sequences, mainly used to assess the tracking capability of algorithms for high-speed moving targets. UAVDT and UAV123, by contrast, provide a broader range of scenarios and target types to enhance the comprehensiveness of the evaluation. These datasets were developed to address the real-time performance and accuracy issues in drone visual tracking, particularly under scenarios with latency and limited computational resources.

- 1Predictive Visual Tracking: A New Benchmark and Baseline Approach同济大学机械工程学院 · 2022年



