Skip to content

Implementation of LKT for tracking objects. Challenges include illumination variance, high frame rate, blurred objects

Notifications You must be signed in to change notification settings

shubham1925/lucas-kanade-template-tracker

Repository files navigation

Lucas Kanade Tracker

Authors

  • Shubham Sonawane
  • Prasheel Renkuntla
  • Raj Prakash Shinde

Description

The project contains the implementation of Lucas Kanade Tracker. The Tracker is tests on three databases viz. Bolt2, Car4 and Dragon Baby. Additionally Huber Loss Function and Histogram Equilization is being implimented to better the results.

Dependencies

  1. Ubuntu 16.04
  2. Python 3.7
  3. openCV 4.2

Output Time

18 mins

Run

  1. To Run tracking of car without huber loss
python3.7 LKT_Car.py
  1. To Run tracking of car with huber loss implementation
python3.7 LKT_Car_Huber.py
  1. To Run LKT on bolt without huber loss approximation
python3.7 lucas_kanade_bolt_without_huber.py
  1. To Run
python3.7 lucas_kanade_bolt_huber_loss.py
  1. To Run tracking of baby face
python3.7 LKT_baby.py 

Results

Image description Image description Image description

About

Implementation of LKT for tracking objects. Challenges include illumination variance, high frame rate, blurred objects

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages