Face Recognition

Implement a face recognition using insightface

These tasks solved with this repo

• Verification: I used InsightFace package for face verification, allowing you to compare two images and determine if they are the same person.

• Identification: I used InsightFace package for face identification, allowing you to recognize different people.

• Face ID: At the end of the readme you will see I wrote a super snake game that it has a face recognition at the beginning of it.

How to install

Run this command:

pip install -r requirements.txt

How to run identification:

First create a face bank directory

face_bank
│
└───Messi
│   │   Messi_image_1.jpg
│   │   Messi_image_2.jpg
│   │   ...
│   
└───Ronaldo
│   │   Ronaldo_image_1.jpg
│   │   Ronaldo_image_2.jpg
│   │   ...
│
│
...

Then run this command to create a face bank

python create_face_bank.py --face-bank ./face_bank

You can also see the other arguments of it with this command

python create_face_bank.py --help

For Example:

• --model: You can chose one of these models. default:buffalo_s
• --face-bank: You can change the face bank directory path. default:face_bank
• --face-bank-name: You can change the face bank file name. default:face_bank.npy

After that run this command

python face_identification.py --img YOUR_IMAGE.JPG

You can also see the other arguments of it with this command

python face_identification.py --help

For Example:

• --model: You can chose one of these models. default:buffalo_s
• --thresh: You can change the threshold of distances between tow faces. default:25
• --face-bank: You can change the face bank file path. default:face_bank.npy
• --update, --no-update: You can update the face bank file. default:False
• --face-bank-path: You can change the face bank directory path if you want to update the face bank file. default:face_bank

Result of identification:

How to run verification:

First run this command to create a face bank

python face_verification.py --img1 YOUR_FIRST_IMAGE.jpg --img2 YOUR_SECOND_IMAGE.jpg

You can also see the other arguments of it with this command

python face_verification.py --help

For Example:

• --model: You can chose one of these models. default:buffalo_s
• --thresh: You can change the threshold of distances between tow faces. default:25
• --show, --no-show: You can see the result by this command default:False

Result of identification:

Face ID in Super Snake:

First make your face bank with create_face_bank.py then run this command:

cd snake \
python main.py

you can change these variables as well:

access, message = FaceIdentification.make_online(
    model='buffalo_s',
    thresh=25,
    face_bank_path='./face_bank.npy',
    camera_id=0,
    is_show=False  # Keep it false I have a threading error here 🥲️
)

Result of face id: