Official repository for AteltaAI Streamer SDK. This repository helps to establish the two video streams across different types of video sources. We are pushing a lot of changes. So keep an update, do star 🌟 the project.
- Download any youtube video from internet, paste the link to get instructed from that video.
- User can also upload private videos for the same.
- Single instructor focussed tracking for multi person videos (BETA)
- Automatic one time pre-processing of the instructor's video.
- Realtime User keypoints detection and tracking to match it with the instructors side by side
- Real time side by side user and instructor video streaming with sound.
Simply running locally
You can run the project by first installing the dependencies by typing:
pip install -r requirements.txt
Then start the app by running:
PYTHONPATH=. streamlit run app/app.py
Building docker image
You can also build the docker image and run the same to run the application. To build the docker image just run
make build_image
Once the image is created, start the app by running:
docker run --privileged --device /dev/video0:/dev/video0 atelta-streamer:latest
This must work on linux distributions and macos. But there might be issues with windows. As there are different ways to give access to camera to the host.
- Real time batched metrics to compare the actions performed by the user and the instructor.
- Metrics for engagement of the user during the session
- Making UI better
- Tracking all the metrics and upload to database
- Video Streamer with gesture/motion controllers for:
- Seeking forward / backward
- Pause
- Resume
- Slow down
- Speed up (Might not required)
Contributing guidelines will be released soon. Till then you can start with pre-existing issues or create new ones. For new issues, We will try to first review and then once approved, you are free to contribute to that.