CDTB
收藏资源简介:
CDTB是由卢布尔雅那大学计算机与信息科学学院和坦佩雷大学信号处理实验室合作创建的大型RGB-D视觉对象跟踪数据集。该数据集通过多种被动和主动RGB-D设置记录,包含室内外序列,特别强调在直接阳光下的户外场景。CDTB是当前最大的RGB-D跟踪数据集,帧数远超相关数据集,旨在为RGB-D跟踪提供一个全面且具有挑战性的基准。数据集中的序列经过精心设计,包含显著的对象姿态变化、杂乱、遮挡和长期目标缺失,以模拟真实跟踪条件。此外,每帧都标注了13个视觉属性,以支持详细的跟踪器性能分析。CDTB的创建旨在推动RGB-D跟踪技术的发展,尤其是在处理复杂场景和对象行为方面的能力。
CDTB is a large-scale RGB-D visual object tracking dataset co-developed by the Faculty of Computer and Information Science, University of Ljubljana and the Signal Processing Laboratory of Tampere University. Recorded using multiple passive and active RGB-D acquisition setups, this dataset encompasses both indoor and outdoor sequences, with a particular focus on outdoor scenes under direct sunlight. Currently the largest RGB-D tracking dataset, CDTB boasts a far greater number of frames than related existing datasets, and is designed to serve as a comprehensive and challenging benchmark for RGB-D object tracking. The sequences in the dataset are meticulously crafted to feature significant object pose variations, clutter, occlusions, and long-term target disappearance, thereby simulating real-world tracking scenarios. Additionally, each frame is annotated with 13 visual attributes to support detailed performance analysis of visual trackers. The development of CDTB aims to advance the state of the art in RGB-D tracking technologies, particularly regarding the capability to handle complex scenes and diverse object behaviors.

- 1CDTB: A Color and Depth Visual Object Tracking Dataset and Benchmark卢布尔雅那大学计算机与信息科学学院 · 2019年



