Object detection and tracking in Python

Part of the upcoming tutorial from my blog poissonisfish featuring an example of my own.

Instructions

• Install Python >= 3.7.0, available here
• Clone this repository to a local / remote machine

Set up virtualenv

From a new terminal, after setting the repository as the working directory:

• Install pipenv by running pip install pipenv

• Run python -m venv <ENV_NAME> to create the pip environment

• Activate the pip environment using one of the following methods:

Platform	Shell	Command to activate virtual environment
POSIX	bash/zsh	$ source <ENV_NAME>/bin/activate
	fish	$ source <ENV_NAME>/bin/activate.fish
	csh/tcsh	$ source <ENV_NAME>/bin/activate.csh
	PowerShell Core	$ /bin/Activate.ps1
Windows	cmd.exe	C:\> <ENV_NAME>\Scripts\activate.bat
	PowerShell	PS C:\><ENV_NAME>\Scripts\Activate.ps1

• Install required packages using pip install -r requirements.txt

Video conversion and object detection

• Create a .MOV video recording and move it to the working directory

• Run ./init.sh <PATH_TO_MOV> to create input and output directories, convert the .MOV video to .mp4 and download the YOLOv3 dependencies for OpenCV. Please note this script requires the following utilities:

  • ffmpeg, available here for Linux, Windows and macOS
  • wget, readily available for Linux; Windows users can download wget.exe here and add the executable path to the system environment PATH variable; macOS users can get it from HomeBrew or similar

• From the active environment, run python tracker.py and watch frame-by-frame computations while the annotated output video is written

Acknowledgements

This work is inspired by a project assignment from the course Computer Vision I at OpenCV.org.