sarahhannes / face-recognition

Face recognition using one shot learning.

Face Recognition

One shot learning using FaceNet and MTCNN.

Quick start

Step 1: Clone repository or download zip

!git clone https://github.com/SarahHannes/face-recognition.git

Step 2: Add training images to /database folder

Step 3: Create a new environment using requirements.txt or face_recog38.yml

Step 4: Activate the created environment on anaconda prompt

conda activate <environment name>

Step 5: Change directory to cloned folder on anaconda prompt

cd <path to cloned folder>

Step 6: Predict! 😌

• To view arguments:
  python main.py --help
• To set path for training folder (required):
  python main.py --database <path to folder containing training images>
• To perform face verification through webcam:
  python main.py --database <path to folder containing training images> --webcam
• To perform face verification on media inputs (.JPG, .MP4):
  python main.py --database <path to folder containing training images> --media <path to folder containing media inputs>
• To specify threshold:
  python main.py --database <path to folder containing training images> -t <value>

References & credits

Thank you!
[1] GitHub repo by foo290
[2] GitHub repo by R4j4n
[3] Shared google drive containing FaceNet weights by Hiroki Taniai
[4] Tutorial on face recognition
[5] Tutorial on argparse