LaST
收藏资源简介:
LaST数据集是由鹏城实验室和北京大学联合创建的大规模时空人物重识别基准,包含10,862个身份和228,156张图像。该数据集通过分析2000多部电影中的场景构建,涵盖了从亚洲到欧洲的多个国家和地区,以及从春季到冬季的不同季节。数据集的创建过程涉及使用半自动标注工具PLabel进行精细标注,确保了数据的高质量和多样性。LaST数据集的应用领域主要集中在人物重识别技术,旨在解决实际场景中由于时空变化导致的人物识别难题,如不同城市、不同时间段和不同服装的人物识别。
The LaST dataset is a large-scale spatiotemporal person re-identification benchmark jointly created by Peng Cheng Laboratory and Peking University, which includes 10,862 identities and 228,156 images. Constructed by analyzing scenes from over 2,000 movies, this dataset covers multiple countries and regions spanning from Asia to Europe, as well as various seasons ranging from spring to winter. The dataset development process involves elaborate annotation using the semi-automatic annotation tool PLabel, ensuring the high quality and diversity of the data. The main application field of the LaST dataset is person re-identification technology, which aims to solve the challenges of person recognition in real-world scenarios caused by spatiotemporal variations, such as person recognition across different cities, time periods and different clothing styles.

- 1Large-Scale Spatio-Temporal Person Re-identification: Algorithms and Benchmark鹏城实验室和北京大学 · 2021年



