Download ViPeD (Train + Validation)
ViPeD is a huge collection of images taken from the highly photo-realistic video game GTA V developed by Rockstar North. It contains automatically generated bounding boxes around every pedestrian so that this dataset can be used to train pedestrian detectors.
The dataset includes a total of about 400K images, split among training and validation subsets. Our GitHub repository contains the code for reading this dataset, together with the code for augmenting the images to emulate some real-world camera effects.
ViPeD extends the JTA (Joint Track Auto) dataset, developed for pose estimation and tracking pursoses.