GRID stands for QMUL underGround Re-IDentification Dataset, which contains 250 pedestrian image pairs, each pair contains two images of the same person (taken by cameras with different perspectives). All images were taken in a subway station with a large flow of people, and were captured by 8 non-adjacent cameras.

The images in this dataset are extremely challenging, with a lot of variations in pose, color, lighting, etc., and the low spatial resolution makes the images very poor quality. The figure next to it shows a snapshot of each camera view of the site and sample images from the dataset.