MOT17 stands for Multiple Object Tracking 17, which is a dataset for multiple object tracking. Similar to the previous version MOT16, the challenges in this dataset include seven indoor and outdoor public place scenes with pedestrians. The video of each scene is divided into two clips, one for training and one for testing.

This dataset detects objects in video frames using three detectors: SDP, Faster-RCNN, and DPM. This challenge accepts both online and offline tracking methods, where the offline tracking method can use future video frames to predict trajectories.