Hyperspectral Tracking Benchmark Dataset
收藏资源简介:
Hyperspectral Tracking Benchmark Dataset是由imec-int机构的研究人员精心构建的大型完全标注超光谱视频数据集,包含35个视频序列,每个序列平均包含500帧,覆盖多种目标类别和复杂场景,如车辆、人脸、动物等。该数据集通过商用高速超光谱相机采集,同时提供与超光谱序列同步的彩色视频,确保了数据的多维性和实用性。数据集的创建旨在推动基于物质信息的视觉目标跟踪研究,特别是在背景杂乱、目标外观快速变化等挑战性场景中的应用。
Hyperspectral Tracking Benchmark Dataset is a large-scale fully annotated hyperspectral video dataset meticulously constructed by researchers from imec-int. It includes 35 video sequences, each with an average of 500 frames, covering diverse target categories and complex scenarios such as vehicles, human faces, animals, and more. The dataset is collected via commercial high-speed hyperspectral cameras, and synchronized color videos aligned with the hyperspectral sequences are also provided, ensuring the multi-dimensionality and practical utility of the data. This dataset was developed to advance visual object tracking research based on material information, particularly for applications in challenging scenarios including cluttered backgrounds and rapid target appearance variations.




