我使用 graph = tf.Graph（）获取图。现在，我使用TensorFlow（更快的Rcnn）创建一个图形。

import cv2
import time
import numpy as np
import core.utils as utils
import TensorFlow as tf
from PIL import Image
return_elements = ["input/input_data:0","pred_sbbox/concat_2:0","pred_mbbox/concat_2:0","pred_lbbox/concat_2:0"]
pb_file         = "./yolov3_coco.pb"
video_path      = "./docs/images/road.mp4"
# video_path      = 0
num_classes     = 80
input_size      = 416
graph           = tf.Graph()
return_tensors  = utils.read_pb_return_tensors(graph,pb_file,return_elements)
with tf.Session(graph=graph) as sess:
    vid = cv2.VideoCapture(video_path)

如何进一步将此图用于物体检测。