How to use CenterNet-52 model inferencing with Camera?

[futureflsl]

Is there a demo about inference from Camera?like this
import cv2
cap = cv2.VideoCapture(0)　　　　#参数为0时调用本地摄像头；url连接调取网络摄像头；文件地址获取本地视频
while(True):
ret,frame=cap.read()
CenterNet_Inference(frame)