/CV_tracking

Primary LanguageJupyter NotebookMIT LicenseMIT

Computer Vision Tracking

This repo contains various Computer Vision tracking projects

Project results

(simple) SORT

Top (No Kalman Filter), Bottom (Kalman Filter)

This video dataset shows the effectiveness of a Kalman Filter with a constant velocity model in the Obstacle Tracking Algorithm.

track_vid_0005_comparison_Trim.mp4

This is a more complex video dataset with pedestrains and vehicles. In this case the constant velocity model in the Kalman Filter is not able to improve tracking and actually degrades it.

track_vid_0028_comparison_Trim.mp4
Dataset sources

https://www.aicrowd.com/challenges/airborne-object-tracking-challenge

http://www.cvlibs.net/datasets/kitti/eval_tracking.php (kitti)