Feelback: An Automated Video Feedback Application

An Automated Video Feedback Application

Feelback is deployed at https://feelbackdemo.me Update: Azure hosting subscription has ended.

---

Video Demo

demo.mp4

Screenshots

System Architecture

---

Install Dependencies

• Install python >= 3.6
• Install conda
• conda env create -f requirements.yml

How to use

CLI

CLI is used only to create annotated video, visualizations, and testing

It DOES NOT generate analytics

Usage

Sample Usage

python feelback_cli.py input_video -o output -f native -v

All Options

Feelback is An Automated Video Feedback Framework

positional arguments:
  input_video           Input Video File Path

options:
  -h, --help            show this help message and exit
  -o filename, --output filename
                        Save processed video to filename.mp4
  --output-annotations annotations [annotations ...]
                        Which annotations to add to processed video
                        annotations can be any of: ['all', 'none', 'ids', 'age', 'gender', 'emotions', 'attention']
                        You can add multiple annotations separated by space
                        Default: all
                        
  --dump filename       Dump Feelback Object After Processing
  --load filename       Load Dumped Feelback Object [For Debug]
  --output-key-moments filename
                        Save Key Moments Visualizations to file
  -f N | native, --fps N | native
                        Process N frames every second, Or `native` to process all frames
  -v, --verbose         Enable more verbosity              

Backend

• Create a .env file like the example.env file and change the following variables: UPLOAD_FOLDER, ANNOTATED_UPLOAD_FOLDER, DATABASE_URL, THUMBNAILS_FOLDER
• run python feelback_server.py to run development server

Frontend

• Install node.js >= 16
• cd into feelback-front
• run npm install
• run npm run serve
• to build for production, run npm run build