Face-Recognition-Surveillance-for-Semi-Public-Spaces

Setup Instructions

1. Download Required Models

• Face Detection 0200
  Download from OpenVINO Model Zoo

• Landmarks Regression Retail 0009
  Download from OpenVINO Model Zoo

• Face Recognition ResNet100 ArcFace (ONNX)
  Download and convert to OpenVINO IR format

2. Install Python Dependencies

pip install -r requirements.txt

3. Run the python file

python retina.py

Test Cases

1. Multiple face detection: detects multiple faces in a single frame without trouble but also mistakenly detects non-facial objects in long-distance and blurred scenarios

2. Face detection in darker lighting: The system can detect multiple faces successfully under low-light conditions.

3. Partially Occluded Face: Face detection is performed successfully even in scenarios involving partially occluded or half-visible faces