Code for visual tracking of objects on ground from a drone.
The objects on the ground are tracked based on the following features:
- Colour : Each object is given a particular color
- Size: The area of the object
- Shape : The objects are further filtered out using a roundness parameter
- Distance between same objects in adjacent frames : This criteria helps to discard tracking of objects which moves beyond a particular distance between the frames
Tested on: Jetson TX1 running on Ubuntu16.04 with an attached ZED camera being carried on a drone.
- OpenCV : Contains the code of the first version of the tracking algorithm
- OpenCV_live : Contains modified code for deploying the algorithm in realtime in a drone
- OpenCV_video : Contains code for testing the algorithm on a pre-recorded video
- Results_Demo : Has demo videos of the algorithm working on pre-recorded videos
How to use these programs:
- Download the repositories
- Open the repository
- Delete the build folder (In the case of OpenCV_video repository make sure to copy the pre-recorded video in the build folder before deleting it)
- Use the following commands to build the code from source file: $ mkdir build $ cd build $ cmake .. $ make
- Use the following command to run the binary file: $ ./ZED_with_OpenCV
NB: The code need to be cleaned up a lot. I will try to do it when I get some free time.