Skip to content

LeoHaoVIP/Single-Object-Tracking

Repository files navigation

Single Object Tracking

This project is an implementation for single object tracking based on mean-shift algorithm, coded in Python language.

Here we use 4 different videos (relatively easy to track) to show the tracking result, most of which are from www.pexels.com, an open community for sharing best free stock videos.

Project Structure

Algorithm Process of Single Object Tracking

  • Calculate the probability histogram of the given target
  • Calculate the probability histogram of the candidate target
  • Find the candidate target which is the closest to the given target with mean-shift

Results Representation

  • Demo Frame

    img_result_1 img_result_2 img_result_3 img_result_4

  • Demo Video

video_result video_result video_result video_result

Dependency

References

  • [1] D. Comaniciu and P. Meer, "Mean shift: a robust approach toward feature space analysis," in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, no. 5, pp. 603-619, May 2002, doi: 10.1109/34.1000236.
  • [2] D. Comaniciu, V. Ramesh and P. Meer, "Kernel-based object tracking," in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 25, no. 5, pp. 564-577, May 2003, doi: 10.1109/TPAMI.2003.1195991.
  • [3] http://cvlab.hanyang.ac.kr/tracker_benchmark/datasets.html

Author Info

LeoHao (XMU-CS)

Date

2020.11.20

About

Python code for single object tracking based on mean-shift

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages