Track Long and Prosper (TLP)

Track Long and Prosper (TLP) 数据集由国际信息技术研究所视觉信息中心创建，包含50个高清视频，总时长超过400分钟，是现有视觉跟踪数据集的20倍以上。该数据集旨在评估长期跟踪性能，支持深度学习模型的训练，减少对数据增强的依赖。数据集涵盖多种目标对象和挑战，如遮挡、快速运动和视角变化，适用于监控、自主导航和增强现实等领域的应用，旨在解决长期跟踪中的挑战，如跟踪算法的一致性和从短暂失败中恢复的能力。

The Track Long and Prosper (TLP) dataset was created by the Visual Information Center of the International Institute of Information Technology. It contains 50 high-definition videos with a total duration exceeding 400 minutes, which is over 20 times that of existing visual tracking datasets. This dataset aims to evaluate long-term tracking performance, support the training of deep learning models, and reduce the reliance on data augmentation. It encompasses a wide range of target objects and challenging scenarios including occlusion, fast motion, and viewpoint variation, and is applicable to applications in fields such as surveillance, autonomous navigation, and augmented reality. Its primary objective is to address core challenges in long-term tracking, such as the consistency of tracking algorithms and the ability to recover from temporary tracking failures.