TrackingNet
收藏资源简介:
TrackingNet是由阿卜杜拉国王科技大学创建的首个大规模野外目标跟踪数据集,包含超过30,000个视频和1400万密集边界框标注。该数据集覆盖了广泛的对象类别和多样化的上下文,旨在训练深度跟踪器并提高其性能和泛化能力。数据集通过从YouTube视频中采样,确保了对象类别的丰富分布,并引入了新的基准测试集,以公平评估未来目标跟踪器的发展。
TrackingNet is the first large-scale in-the-wild object tracking dataset created by King Abdullah University of Science and Technology (KAUST). It contains over 30,000 videos and 14 million densely annotated bounding boxes. This dataset covers a wide range of object categories and diverse contextual environments, aiming to train deep trackers and improve their performance and generalization capabilities. Developed by sampling from YouTube videos, it ensures a rich distribution of object categories, and introduces a novel benchmark to fairly evaluate the advancement of future object trackers.




