this project is implemented based on the python implementation of uoip from https://github.com/uoip/KCFpy
Documents, videos, demos and some other results can be founded in following google drive link: https://drive.google.com/drive/folders/0B1cwJfdr4A45WVdTZzhiMkk0RWM?usp=sharing
In this repo,
- /DesignReview includes the reports
- /kcf_py includes the src of the project
- /kcf_master a c++ version kcf.
**run kcf_py/run.py to run the program
- Usage: python run.py will connect to the camera, and python run.py filename will use video file as input
- by default using dlib detector. Install RCNN lib and modify kcftracker.py line 7 & 109 to use RCNN dector.
- check run.py line 34 for hog features
- check branch: mix-ranking for an alternative score function.
- check run.py line 94-104 to show trajectory.
Contact: leonhuxf@umich.edu