Caltech Roadside Pedestrians (CRP)
收藏资源简介:
Caltech Roadside Pedestrians (CRP)数据集由加州理工学院创建,旨在研究行人细粒度分类。该数据集包含27,454个边界框和姿态标签,记录了来自移动车辆的自然环境下行人的视频。数据集内容丰富,包括年龄、性别、体重和服装风格等多个细粒度分类。创建过程中,使用Amazon Mechanical Turk进行标注,确保了数据的高质量和详细性。该数据集适用于训练和测试细粒度分类算法,也支持行人跟踪、检测和姿态估计的基准测试。
Caltech Roadside Pedestrians (CRP) dataset was developed by the California Institute of Technology for research on fine-grained pedestrian classification. This dataset contains 27,454 bounding boxes and pose labels, recording videos of pedestrians in natural environments captured from moving vehicles. It includes rich fine-grained classification attributes such as age, gender, weight, clothing style and others. During the dataset creation process, Amazon Mechanical Turk was employed for annotation to guarantee high data quality and detailed labeling. This dataset is applicable for training and testing fine-grained classification algorithms, and also supports benchmark tests for pedestrian tracking, detection and pose estimation.

- 1Fine-Grained Classification of Pedestrians in Video: Benchmark and State of the Art加州理工学院 · 2016年



