# intercept stream and inference on video

**URL:** <https://community.wowza.com/t/intercept-stream-and-inference-on-video/55558>\
**Category:** Hire a Consultant\
**Tags:** wowza-streaming-server-java-api\
**Created:** [July 15, 2020, 5:21pm UTC](https://community.wowza.com/t/intercept-stream-and-inference-on-video/55558 "2020-07-15T17:21:33Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Ajander\_Singh](https://avatars.discourse-cdn.com/v4/letter/a/9fc348/32.png) [@Ajander\_Singh](https://community.wowza.com/u/Ajander_Singh)\
**Post date:** [July 15, 2020, 5:21pm UTC](https://community.wowza.com/t/intercept-stream-and-inference-on-video/55558/1 "2020-07-15T17:21:33Z")

</div>

Hi,

I have a requirement to apply analytics on received stream and restream after inferencing on stream/video. So I wan to intercept stream and apply AI to detect object and draw bounding box on stream/image and re-encode to restream for subscriber. Subscriber can see video with object detected bounding box. I should be able to extract audio from video and save Audio and video in MP4 format on file system or in Azure cloud.

Sorry, I am new to media server so please bear with me.

Please suggest me how can I do that and would I have to create new Wowza project/module or what will be best approach for the same.

Thanks,

---

<div class="post-metadata">

**Author:** ![Cassidy\_Reilly](https://avatars.discourse-cdn.com/v4/letter/c/c67d28/32.png) [@Cassidy\_Reilly](https://community.wowza.com/u/Cassidy_Reilly)\
**Post date:** [July 15, 2020, 7:02pm UTC](https://community.wowza.com/t/intercept-stream-and-inference-on-video/55558/2 "2020-07-15T19:02:38Z")

</div>

That’s a pretty specific, advanced workflow so I would suggest submitting a [support ticket](https://www.wowza.com/community/spaces/26/find-a-consultant.html) to see if Wowza could assist you in building that or you can also try posting in [Hire A Consultant](https://www.wowza.com/community/spaces/26/find-a-consultant.html) here in the forums. That will indeed require some custom work and we have no sample code on that unfortunately.
