Multiple Object: Tracker 2D (Python API)

• Introduction
• Running Software
• Requirements
• Status
• Related projects

Introduction

multipleObjectTracker2D module use OpenCV Python API. This module tracks objects, people, animals or something that the user needs in a image source. multipleObjectTracker2D use OpenCV region of interest to select the region or different regions that the user need to track selecting with a rectangle with user mouse input.

multipleObjectTracker2D is a multipurpose tracker, that support different OpenCV trackers like CSRT, GOTURN, KCF ...

The module can be configured in config.ini file, selecting the track you want to use, also image source size to be resize and improve speed. By default use user webcam, but can be configured to use IP video source, local file or OpenCV supported sources.

Also this module has support to use YARP middleware, and can receive video source from YARP port, the module receive in /multipleObjectTracker2D/img:i port. Also publish in YARP network in /multipleObjectTracker2D/img:o processed image source.

Centroid tracked object coordinates is published in /multipleObjectTracker2D/data:o YARP port.

The module can be used with or without YARP dependences and configured in config.ini file.

Documentation available on docs.

Running Software

• Running without YARP support

1. Execute programs/multipleObjectTracker2D.py the detector.
2. Select region to track and q to end selection or c to continue selecting multiple object to track.

• Running with YARP support

1. Configure config.ini to select yarp-send and yarp-receive if you want to use YARP input, output or both of them.
2. Create or configure YARP Server.
3. Connect YARP input port if needed.

yarp connect /videoSource /multipleObjectTracker2D/img:i

4. Execute programs/multipleObjectTracker2D.py the detector.
5. Select region to track and q to end selection or c to continue selecting multiple object to track.

NOTE:

• Video results are published on /multipleObjectTracker2D/img:o
• Coordinate results are published on /multipleObjectTracker2D/data:o

Requirements

multipleObjectTracker2D requires:

• Install pip3
• Install OpenCV and OpenCV Contrib

pip3 install opencv-python==3.4.7.28
pip3 install opencv-contrib-python==3.4.7.28

• Install YARP 2.3.XX+ with Python 3 bindings

Tested on: windows 10, ubuntu 16.04, ubuntu 18.04, lubuntu 18.04 and kubuntu 20.04.

Status

Related projects

• OpenCV: Tracking API
• OpenCV: Introducction to Tracking
• davidvelascogarcia: Tensorflow: Detector 2D (Python API)