CCGR
收藏资源简介:
CCGR是由中南大学和南方科技大学联合创建的大型步态识别数据集,包含约160万条序列,覆盖970个主题。每个主题具有33个视角和53种不同的协变量,提供了RGB、解析、轮廓和姿态等多种类型的步态数据。数据集通过20个月的努力收集而成,旨在解决现有数据集在协变量多样性方面的不足。CCGR适用于深入研究跨协变量步态识别,特别是在实际应用中面临的挑战。
CCGR is a large-scale gait recognition dataset jointly created by Central South University and Southern University of Science and Technology. It contains approximately 1.6 million sequences covering 970 subjects. Each subject has 33 viewpoints and 53 different covariates, and the dataset provides multiple types of gait data including RGB, segmentation, silhouette and pose data. The dataset was collected over a 20-month period, aiming to address the insufficient covariate diversity of existing datasets. CCGR is suitable for in-depth research on cross-covariate gait recognition, especially the challenges faced in practical applications.

- 1Cross-Covariate Gait Recognition: A Benchmark中南大学自动化学院 南方科技大学计算机科学与工程系 南方科技大学可信自主系统研究院 香港大学 · 2024年



