EmojiPrivacy

A simple python script to overlay faces in a video with emojis for privacy. Additionally the facial expression gets analysed so that the expression gets projected onto one of seven emojis (angry, disgust, fear, happy, neutral, sad and surprise)

Requirements

pip install tensorflow deepface

Usage

python process_video.py <input_video> [--output_video] [--emoji_scaler]

• input_video: Path to the input video file. This is a required argument.
• output_video: Path to the output video file. This is an optional argument. If not provided, the output filename will be the same as the input filename, appended with _output.
• emoji_scaler: A floating-point value to scale the size of the emojis. This is an optional argument. If not provided, it defaults to 1.5.

Example Usages

python process_video.py input_video.mp4
python process_video.py input_video.mp4 --output_video output_video.mp4 --emoji_scaler 2.0

Output Test

Original Video: https://www.youtube.com/watch?v=ydAyvvDQrgY

Limitations

Currently no audio of the input video gets copied and there is no possibility to exclude specific faces. I am currently working on the "exclude face feature".

License

This project is licensed under the terms of the Apache License 2.0.

For details, see the LICENSE file.