App developed to assist in detecting issues with public performances and speeches. After uploading an .mp4 video file, the app will analyze the speech and provide a feedback on the quality of the speech, as well as the potential issues.
Software stack consists of a Flask backend connected to a custom job scheduler (app/job_manager
) running media procesing pipelines concurently. All analysis modules placed in the app/modules
submodule and are designed in an modular fashion. Pipeline description is located at job.yaml
. In future the job manager component processing throughput could be improved by spinning more instances. The frontend is based on React framework allowing end user to interact with analyzed data.
.
Developed during the HackYeah 2024 hackathon.
Repo setup:
git clone https://github.com/spejsiaczki/hy_app.git
cd hy_app
git submodule update --init --recursive
Python dependencies:
python -m venv .venv
source .venv/bin/activate
pip install -r app/modules/requirements.txt
Build the app:
cd web
npm install
npm run build
cd ..
Start the app:
cd app
flask run
Additionaly a Docker container recipe is provided, app by default will listen the 5000
port.