release_data.zip

1. This is the README file for PA-100K dataset. 2. The 100,000 pedestrian images are in folder data/. 3. The annotation information is in annotation.mat. The variables in this file are listed as follows: "train_images_name", "val_images_name" and "test_images_name" are names of training, validation and test images respectively. "train_label", "val_label" and "test_label" are labels of images samples. It is of size n x 26, corresponding to n images in the corresponding set and 26 attributes. "attributes" is a 26 x 1 cell, showing the 26 attributes' names. 4. If you use the dataset, please cite this paper: