import cv2 as cv

# 加载模型
net = cv.dnn.readNet('/home/kang/open_model_zoo-2019/model_downloader/Transportation/object_detection/face/pruned_mobilenet_reduced_ssd_shared_weights/dldt/face-detection-adas-0001.xml', '/home/kang/open_model_zoo-2019/model_downloader/Transportation/object_detection/face/pruned_mobilenet_reduced_ssd_shared_weights/dldt/face-detection-adas-0001.bin')

# 指定目标设备CPU
net.setPreferableTarget(cv.dnn.DNN_TARGET_CPU)

# 加载图像
frame = cv.imread('/home/kang/Desktop/0903171727.png')

# 准备输入 blob 
blob = cv.dnn.blobFromImage(frame, size=(672, 384), ddepth=cv.CV_8U)

# 执行推理
net.setInput(blob)
out = net.forward()

# 绘制检测到的脸
for detection in out.reshape(-1, 7):

    confidence = float(detection[2])

    xmin = int(detection[3] * frame.shape[1])
    ymin = int(detection[4] * frame.shape[0])
    xmax = int(detection[5] * frame.shape[1])
    ymax = int(detection[6] * frame.shape[0])
    # BGR格式
    # if confidence > 0.5:
    # 为了识别更多的人脸，可以调节检测的置信度，使得程序对人脸的判定没那么严格，这里下调为0.2(原来是0.5)
    if confidence > 0.2:
        cv.rectangle(frame, (xmin, ymin), (xmax, ymax), color=(0, 255, 0))

# 保存图像文件
cv.imwrite('./out.png', frame)