This repo illustrates server sending a video of a bouncing ball to client and client responding with the predicted location of the moving ball using OpenCV and aiortc.
First ensure the correct Python version (python3.10):
$ python -V
Second install the required packages:
$ pip install -r requirements.txt
To send offer to client:
$ python3 server.py
Once the offer signal from server is established, run client.py to process the video:
$ python3 client.py
The client will display the frames of the green moving ball with the red contour enclosing the object. The server will receive answer containing the predicted location of the ball from client, and compute the percentage error of the predicted and true values.
lite_demo.mp4
Check BallVideoStreamTrack() object attributes and localization results threshold:
test_media.py -- ensure ball radius, (x,y) coordinates, screen width and heights, etc are within constraints
test_localize.py -- ensure localization percentage error of video frame are less than 10%
From the project folder, run python3 -m pytest tests or python3 -m pytest tests -W ignore::DeprecationWarning to ignore deprecation warnings
Makefile for Docker and Kubernetes commands:
To build server image $ make build-server-image
To build client image $ make build-client-image
To run server $ make docker-run-server
To run client $ make docker-run-client
To deploy server $ make k8s-deploy-server
To deploy client $ make k8s-deploy-client
Template: https://github.com/aiortc/aiortc/blob/main/examples/server/README.rst