# -*-coding:utf-8-*-
'''
DpCas-Light
||||      |||||        ||||         ||       |||||||
||  ||    ||   ||    ||    ||      ||||     ||     ||
||    ||  ||    ||  ||      ||    ||  ||     ||
||    ||  ||   ||   ||           ||====||     ||||||
||    ||  |||||     ||      ||  ||======||         ||
||  ||    ||         ||    ||  ||        ||  ||     ||
||||      ||           ||||   ||          ||  |||||||

/--------------------- Face Bioassay ---------------------/
'''
# date:2021-04-18
# Author: Eric.Lee
# function: Face Bioassay "基于人脸动作的活体检测"

import cv2

# 加载模型组件库
from components.face_detect.yolo_v3_face import yolo_v3_face_model
from components.face_multi_task.face_multi_task_component import FaceMuitiTask_Model
from components.face_euler_angle.face_euler_angle_component import FaceAngle_Model
from components.face_bioassay.face_bioassy_component import FaceBioassyModel
# 加载工具库
from lib.face_bioassay_lib.cores.face_fuction import get_faces_batch_attribute
from lib.face_bioassay_lib.utils.utils import parse_data_cfg


def main_face_bioassay(video_path, cfg_file):
    config = parse_data_cfg(cfg_file)
    print("\n/************** Face Bioassay *****************/")
    print("/********************************************************/\n")
    # pose_model = light_pose_model()
    face_detect_model = yolo_v3_face_model(conf_thres=float(config["detect_conf_thres"]),
                                           nms_thres=float(config["detect_nms_thres"]),
                                           model_arch=config["detect_model_arch"],
                                           model_path=config["detect_model_path"],
                                           yolo_anchor_scale=float(config["yolo_anchor_scale"]),
                                           img_size=float(config["detect_input_size"]),
                                           )
    face_multitask_model = FaceMuitiTask_Model(model_path=config["face_multitask_model_path"],
                                               model_arch=config["face_multitask_model_arch"])
    face_euler_model = FaceAngle_Model(model_path=config["face_euler_model_path"])
    face_bioassy_model = FaceBioassyModel(model_path='')

    cap = cv2.VideoCapture(video_path)
    frame_idx = 0
    while cap.isOpened():
        ret, img = cap.read()
        if ret:
            frame_idx += 1
            video_time = cap.get(cv2.CAP_PROP_POS_MSEC)

            faces_bboxes = face_detect_model.predict(img, vis=False)  # 检测手，获取手的边界框

            faces_message = get_faces_batch_attribute(
                face_multitask_model, face_euler_model, face_bioassy_model, faces_bboxes, img, use_cuda=True, vis=True)
            if faces_message is not None:
                print("faces_message : {} \n".format(faces_message))

            cv2.namedWindow("DriverFatigueMonitor", 0)
            cv2.imshow("DriverFatigueMonitor", img)
            if cv2.waitKey(1) == 27:
                break
        else:
            break

    cv2.destroyAllWindows()