MIAP Pedestrian Detection Image Dataset

MIAP stands for More Inclusive Annotations for People. It is an extension of a subset of the Open Images dataset, containing 100,000 images (70,000 from the training set and 30,000 from the test/validation set). All visible people in the images are annotated with bounding boxes and attributes, including gender (mainly female, male, and unknown) and age range (young, middle-aged, old, and unknown). The image annotations in this dataset are more detailed than the original dataset, with nearly 100,000 bounding boxes added. It is mainly used for machine learning fairness research (ML Research).