QuadTrack600
收藏资源简介:
QuadTrack600是一个包含RGB、热红外、事件和语言四种模态数据的大型高质量基准数据集,由安徽大学构建,旨在为通用多模态视觉跟踪提供一个全面的评估平台。该数据集包含600个视频序列,每个序列包含三种模态的数据以及针对第一个帧的语言描述,总计超过73万个帧对。每个帧对在空间上对齐,并手动标注边界框,确保高质量注释。该数据集在行人、交通工具和动物等多种目标类别上具有高度多样性,适用于评估多模态视觉跟踪方法在处理复杂场景时的性能。
QuadTrack600 is a large-scale high-quality benchmark dataset containing four modalities: RGB, thermal infrared, event, and language. It was constructed by Anhui University, aiming to provide a comprehensive evaluation platform for general-purpose multimodal visual tracking. This dataset includes 600 video sequences, each with three modalities of data and a language description for the first frame, with a total of over 730,000 frame pairs. Each frame pair is spatially aligned and manually annotated with bounding boxes to ensure high-quality annotations. The dataset boasts high diversity across multiple target categories such as pedestrians, vehicles and animals, and is suitable for evaluating the performance of multimodal visual tracking methods when dealing with complex scenarios.

- 1Towards General Multimodal Visual Tracking安徽大学计算机科学与技术学院,安徽大学人工智能学院 · 2025年



