import cv2
import face_recognition

# 初始化摄像头
cap = cv2.VideoCapture(0)

# 调整摄像头捕捉的图像大小
cap.set(3, 640)  # 宽度
cap.set(4, 480)  # 高度

# 加载已知人脸图像并提取特征
known_face_img = face_recognition.load_image_file(r"C:\Users\linyiwu\Desktop\datasets\face\train\0002.jpg")
known_face_encoding = face_recognition.face_encodings(known_face_img)[0]

while True:
    ret, frame = cap.read()

    # 将图像转换为 RGB 格式（face_recognition 要求输入为 RGB 格式）
    rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)

    # 在图像中检测人脸
    face_locations = face_recognition.face_locations(rgb_frame)
    face_encodings = face_recognition.face_encodings(rgb_frame, face_locations)

    # 对检测到的人脸进行处理
    for (top, right, bottom, left), face_encoding in zip(face_locations, face_encodings):
        # 进行人脸识别
        matches = face_recognition.compare_faces([known_face_encoding], face_encoding)
        name = "Unknown"

        if matches[0]:
            name = "Known Person"

        # 在人脸周围绘制矩形并显示姓名
        cv2.rectangle(frame, (left, top), (right, bottom), (0, 255, 0), 2)
        cv2.putText(frame, name, (left + 6, bottom - 6), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 1)

    # 显示图像
    cv2.imshow('Face Recognition', frame)

    # 按下 'q' 键退出循环
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

# 释放摄像头资源并关闭窗口
cap.release()
cv2.destroyAllWindows()

'''
Author: lin
Time: 2024年4月22日10:40:22
Description: 实现摄像头的调用，人脸检测和验证的合并
'''
import cv2
from deepface import DeepFace

# 初始化摄像头
cap = cv2.VideoCapture(0)

# 调整摄像头捕捉的图像大小
cap.set(3, 640)  # 宽度
cap.set(4, 480)  # 高度

# 设置用于比对的人脸图像文件夹路径
base_folder = r'C:\Users\linyiwu\Desktop\datasets\face\train'
img2_pathe = r"C:\Users\linyiwu\Desktop\datasets\face\train\0001.jpg"

while True:
    ret, frame = cap.read()

    # 保存摄像头捕捉的图像到临时文件
    temp_face_path = r"C:\Users\linyiwu\Desktop\datasets\face\train\temp_face.jpg"
    cv2.imwrite(temp_face_path, frame)

    # 调用DeepFace进行人脸验证
    try:
        # result = DeepFace.verify(img1_path=temp_face_path, img2_path=img2_path, model_name='GhostFaceNet', enforce_detection=False)
        # Debug
        # result = DeepFace.verify(img1_path=img2_pathe, img2_path=img2_pathe, model_name='DeepFace', enforce_detection=False)
        result = DeepFace.verify(img1_path=r'C:\Users\linyiwu\Desktop\datasets\face\train\temp_face.jpg', img2_path=r"C:\Users\linyiwu\Desktop\datasets\face\train\0001.jpg")

        # 提取验证结果
        verified = result['verified']
        distance = result['distance']

        if verified:
            verified_text = f"Verified: {verified} - distance: {distance:0.4}"
        else:
            verified_text = f"Not Verified - distance: {distance:0.4}"

        cv2.putText(frame, verified_text, (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 0.9, (0, 255, 0), 2)

        # 显示图像
        cv2.imshow('Face Verification', frame)
    except:
        cv2.imshow('Face Verification', frame)
        print("No face detected")
    # 按下 'q' 键退出循环
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

# 释放摄像头资源并关闭窗口
cap.release()
cv2.destroyAllWindows()