Velocity Calculation with Kinectv2 Camera

This repository is the location of Group IV's final project for CS 549, Fall 2022. The following project aims to estimate the velocity of a detected object using YOLOv3, norfair, and PyKinect2. YOLOv3 was published in research paper: YOLOv3: An Incremental Improvement: Joseph Redmon, Ali Farhadi It's originally implemented in YOLOv3.

BEFORE RUNNING!

Please edit the following variables in the User Inputs section at the top of the pykinect2_norfair_merge.py file:

• path - Location of the cfg and weights folders that contain the .cfg and .weights files for YOLOv3-tiny and YOLOv3.
• tiny - bool: True if YOLOv3-tiny is to be used for the object detection; False if YOLOv3 is to be used.
• object_of_interest - Change this based on the object that's velocity you would like to track. NOTE: name must be in coco.names.
• n - number of frames to average the velocity over. This average velocity value is what is displayed in the live stream.

Once all user inputs have been configured, run:

pykinect2_norfair_merge.py

Z Velocity Testing with Soccer Ball and YOLOv3

Z_velocity_test_soccer_ball.mp4

Requirements

• Python 3.7.9
• OpenCV 4.2.0
• YOLOv3
• PyKinect2

Dependencies

• opencv
• numpy
• norfair