deface with Selective Face Blurring for Lecture Videos

This project is a modified version of the deface repository, designed specifically for treating lecture videos to protect student privacy while keeping the professor's face visible.

Features

• Selective Face Blurring: At the beginning of a video, the professor's face can be selected to remain unblurred while all other faces are blurred throughout the lecture.
• Privacy Preservation: Helps anonymize student faces in educational videos, ideal for online lecture recordings.

Usage

The CLI usage remains the same as the original Deface repository. For details on how to run the tool, refer to the Deface documentation.

Example Command

python deface/main.py /path/to/lecture.mp4

In this modified version, a face is selected at the start of the video for exclusion from the blurring process.

Other Experiments

• Tested face_recognition, EgoBlur, OpenCV Multi-Object Tracking API and Youtube Studio (in-built blur) to blur student faces. All fell short of providing a high accuracy consistent blur.

Credit

Project idea from Divide-By-0 and MIT SOUL