RAP Pedestrian Attribute Recognition Dataset

RAP stands for Richly Annotated Pedestrian, which is a dataset for pedestrian attribute recognition. It contains 41,585 images collected from indoor surveillance cameras, each image is annotated with 72 attributes, and only 51 binary attributes with a positive rate higher than 1% are selected for evaluation. The training set of this dataset has 33,268 images and the test set has 8,317 images.