avinash793/object-tracking-software

Implementation of Lucas Kanade Tracking system using six parameter affine model and recursive Gauss-Newton process.

Object Tracking

Track moving object in a given video.

Requirements

• Numpy >= 1.24.3
• OpenCV >= 4.9.0 (latest as of 2024)

Description

We have implemented Object Tracking Software which tracks moving object in a given video. We have implemented in two ways:

1. From scratch using Lucas Kanade Tracking algorithm which uses six parameter affine model and recursive Gauss-Newton algorithm. Implemented the research paper Lucas-Kanade 20 Years On: by simon Baker (Microsoft Computer vision researcher).
2. using OpenCV library.

Then, We have done the detailed comparative analysis of both implementation which can be found in the attached pdf.

How To Run

1. Put video in video/ folder, or you can use provided sample video slow_traffic.mp4 .
2. Run:

   python3 lucas_kanade_scratch.py

   or

   python3 lucas_kanade_opencv.py

3. Enter video filename along with path like:

   Enter video filename along with path: video/slow_traffic.mp4

   Enter "0" to use your real time system camera for video.

object-tracking-software's Issues

anyway i can define multiple ROI instead of using feature points?