from_camera.py

# Tutorial: https://youtu.be/PdkPI92KSIs?list=PLQVvvaa0QuDcDqgpLLJJM15NpIGNfrKY5

import os

import cv2
import face_recognition

KNOWN_FACES_DIR = "known_faces"
TOLERANCE = 0.5
FRAME_THICKNESS = 1
FONT_THICKNESS = 1
MODEL = "hog"

video = cv2.VideoCapture(0)

print("[*] Loading known faces...")

known_faces = []
known_names = []

for name in os.listdir(KNOWN_FACES_DIR):
    for filename in os.listdir(f"{KNOWN_FACES_DIR}/{name}"):
        image = face_recognition.load_image_file(f"{KNOWN_FACES_DIR}/{name}/{filename}")
        encoding = face_recognition.face_encodings(image)[0]
        known_faces.append(encoding)
        known_names.append(name)

print("[*] Processing camera feed...")

while True:
    ret, image = video.read()

    locations = face_recognition.face_locations(image, model=MODEL)
    encodings = face_recognition.face_encodings(image, locations)

    for face_encoding, face_location in zip(encodings, locations):
        results = face_recognition.compare_faces(known_faces, face_encoding, TOLERANCE)
        match = None
        if True in results:
            match = known_names[results.index(True)]

            top_left = (face_location[3], face_location[0])
            bottom_right = (face_location[1], face_location[2])
            color = [0, 255, 0]
            cv2.rectangle(image, top_left, bottom_right, color, FRAME_THICKNESS)

            top_left = (face_location[3], face_location[2])
            bottom_right = (face_location[1], face_location[2] + 22)
            cv2.rectangle(image, top_left, bottom_right, color, cv2.FILLED)
            cv2.putText(
                image,
                match,
                (face_location[3] + 10, face_location[2] + 15),
                cv2.FONT_HERSHEY_TRIPLEX,
                0.35,
                (0, 0, 0),
                FONT_THICKNESS,
            )

    cv2.imshow(filename, image)
    if cv2.waitKey(1) & 0xFF == ord("q"):
        break