OpenCV ObjDetect Module Face Recognition (SFace) Sample

README.md

OpenCV ObjDetect Module Face Recognition (SFace) Sample

• 1. Generate Aligned Faces

  python generate_aligned_faces.py image.jpg
  

  • Input
    

  • Outputs
    
    

• 2. Generate Feature Dictionary

  python generate_feature_dictionary.py face001.jpg
  python generate_feature_dictionary.py face002.jpg
  

  • Inputs
    
    

  • Outputs

    • face001.npy
    • face002.npy

• Face Recognizer

  • Input
    

  • Output (NOTE: label is file name of dictionary)
    

Download

yunet.onnx
face_recognizer_fast.onnx

Reference

DNN-based Face Detection And Recognition | OpenCV Tutorials
cv::FaceDetectorYN Class Reference | OpenCV Online Documentation
cv::FaceRecognizerSF Class Reference | OpenCV Online Documentation

Update

2023/06/27 OpenCV DNN based Face Detection API/Model will update to YuNet v2 from OpenCV 4.8.0. If you want more detail, Please refer to Pull Request.

face_recognizer.py

import os
import sys
import glob
import numpy as np
import cv2

COSINE_THRESHOLD = 0.363
NORML2_THRESHOLD = 1.128

# 特徴を辞書と比較してマッチしたユーザーとスコアを返す関数
def match(recognizer, feature1, dictionary):
    for element in dictionary:
        user_id, feature2 = element
        score = recognizer.match(feature1, feature2, cv2.FaceRecognizerSF_FR_COSINE)
        if score > COSINE_THRESHOLD:
            return True, (user_id, cos_score)
    return False, ("", 0.0)

def main():
    # キャプチャを開く
    directory = os.path.dirname(__file__)
    capture = cv2.VideoCapture(os.path.join(directory, "image.jpg")) # 画像ファイル
    #capture = cv2.VideoCapture(0) # カメラ
    if not capture.isOpened():
        exit()

    # 特徴を読み込む
    dictionary = []
    files = glob.glob(os.path.join(directory, "*.npy"))
    for file in files:
        feature = np.load(file)
        user_id = os.path.splitext(os.path.basename(file))[0]
        dictionary.append((user_id, feature))

    # モデルを読み込む
    weights = os.path.join(directory, "yunet.onnx")
    face_detector = cv2.FaceDetectorYN_create(weights, "", (0, 0))
    weights = os.path.join(directory, "face_recognizer_fast.onnx")
    face_recognizer = cv2.FaceRecognizerSF_create(weights, "")

    while True:
        # フレームをキャプチャして画像を読み込む
        result, image = capture.read()
        if result is False:
            cv2.waitKey(0)
            break

        # 画像が3チャンネル以外の場合は3チャンネルに変換する
        channels = 1 if len(image.shape) == 2 else image.shape[2]
        if channels == 1:
            image = cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)
        if channels == 4:
            image = cv2.cvtColor(image, cv2.COLOR_BGRA2BGR)

        # 入力サイズを指定する
        height, width, _ = image.shape
        face_detector.setInputSize((width, height))

        # 顔を検出する
        result, faces = face_detector.detect(image)
        faces = faces if faces is not None else []

        for face in faces:
            # 顔を切り抜き特徴を抽出する
            aligned_face = face_recognizer.alignCrop(image, face)
            feature = face_recognizer.feature(aligned_face)

            # 辞書とマッチングする
            result, user = match(face_recognizer, feature, dictionary)

            # 顔のバウンディングボックスを描画する
            box = list(map(int, face[:4]))
            color = (0, 255, 0) if result else (0, 0, 255)
            thickness = 2
            cv2.rectangle(image, box, color, thickness, cv2.LINE_AA)

            # 認識の結果を描画する
            id, score = user if result else ("unknown", 0.0)
            text = "{0} ({1:.2f})".format(id, score)
            position = (box[0], box[1] - 10)
            font = cv2.FONT_HERSHEY_SIMPLEX
            scale = 0.6
            cv2.putText(image, text, position, font, scale, color, thickness, cv2.LINE_AA)

        # 画像を表示する
        cv2.imshow("face recognition", image)
        key = cv2.waitKey(1)
        if key == ord('q'):
            break
    
    cv2.destroyAllWindows()

if __name__ == '__main__':
    main()

generate_aligned_faces.py

import os
import argparse
import numpy as np
import cv2

def main():
    # 引数をパースする
    parser = argparse.ArgumentParser("generate aligned face images from an image")
    parser.add_argument("image", help="input image file path (./image.jpg)")
    args = parser.parse_args()

    # 引数から画像ファイルのパスを取得
    path = args.image
    directory = os.path.dirname(args.image)
    if not directory:
        directory = os.path.dirname(__file__)
        path = os.path.join(directory, args.image)

    # 画像を開く
    image = cv2.imread(path)
    if image is None:
        exit()

    # 画像が3チャンネル以外の場合は3チャンネルに変換する
    channels = 1 if len(image.shape) == 2 else image.shape[2]
    if channels == 1:
        image = cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)
    if channels == 4:
        image = cv2.cvtColor(image, cv2.COLOR_BGRA2BGR)

    # モデルを読み込む
    weights = os.path.join(directory, "yunet.onnx")
    face_detector = cv2.FaceDetectorYN_create(weights, "", (0, 0))
    weights = os.path.join(directory, "face_recognizer_fast.onnx")
    face_recognizer = cv2.FaceRecognizerSF_create(weights, "")

    # 入力サイズを指定する
    height, width, _ = image.shape
    face_detector.setInputSize((width, height))

    # 顔を検出する
    _, faces = face_detector.detect(image)

    # 検出された顔を切り抜く
    aligned_faces = []
    if faces is not None:
        for face in faces:
            aligned_face = face_recognizer.alignCrop(image, face)
            aligned_faces.append(aligned_face)

    # 画像を表示、保存する
    for i, aligned_face in enumerate(aligned_faces):
        cv2.imshow("aligned_face {:03}".format(i + 1), aligned_face)
        cv2.imwrite(os.path.join(directory, "face{:03}.jpg".format(i + 1)), aligned_face)

    cv2.waitKey(0)
    cv2.destroyAllWindows()

if __name__ == '__main__':
    main()

generate_feature_dictionary.py

import os
import sys
import argparse
import numpy as np
import cv2

def main():
    # 引数をパースする
    parser = argparse.ArgumentParser("generate face feature dictionary from an face image")
    parser.add_argument("image", help="input face image file path (./face.jpg)")
    args = parser.parse_args()
    print(args.image)

    # 引数から画像ファイルのパスを取得
    path = args.image
    directory = os.path.dirname(args.image)
    if not directory:
        directory = os.path.dirname(__file__)
        path = os.path.join(directory, args.image)

    # 画像を開く
    image = cv2.imread(path)
    if image is None:
        exit()

    # 画像が3チャンネル以外の場合は3チャンネルに変換する
    channels = 1 if len(image.shape) == 2 else image.shape[2]
    if channels == 1:
        image = cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)
    if channels == 4:
        image = cv2.cvtColor(image, cv2.COLOR_BGRA2BGR)

    # モデルを読み込む
    weights = os.path.join(directory, "face_recognizer_fast.onnx")
    face_recognizer = cv2.FaceRecognizerSF_create(weights, "")

    # 特徴を抽出する
    face_feature = face_recognizer.feature(image)
    print(face_feature)
    print(type(face_feature))

    # 特徴を保存する
    basename = os.path.splitext(os.path.basename(args.image))[0]
    dictionary = os.path.join(directory, basename)
    np.save(dictionary, face_feature)

if __name__ == '__main__':
    main()

return True, (user_id, cos_score)
は
return True, (user_id, score)
と置き換えると、face_recognizer.py は動作しました。