0

I've used the Video Intelligence API to do object tracking on my video - and have received a JSON output with all the frames and objects detected in my video.

Is there a way I can generate a video output from my JSON, using the original video - to show an example of all the objects tracked for the video?

Something like the below:

https://cdn-images-1.medium.com/max/500/1*q1uVc-MU-tC-WwFp2yXJow.gif

re0603
  • 387
  • 1
  • 4
  • 19

1 Answers1

0

Video Intelligence API does not generate any video where objects are shown to be tracked with drawn boxes on objects that are detected as shown in your example link. You have to do this separately or manually to images/videos and use other image annotation tool.

You can refer to this article from Spring ML which explains how you can achieve such output. The example use case discussed was about detecting potholes along the road which used an image annotation tool. With the kind of video output you are looking for, you need to work a lot in order to achieve this. Here is a brief summary about the article.

  1. Spliced the video into image/video frames.
  2. Train your own model to detect desired object of interest.
  3. Use Label IMG to identify and label objects of interest.
  4. Apply the labelled images/videos.
DavicC
  • 96
  • 5