Run a face detection process on live stream

Question

Greetings, I am currently trying to create an service/app to detect people from a collection on a live streaming or a feed from a camera and I want to be able to notify a consumer app that we have detected a know person and maybe show a clip but, reading the docs and doing some testing I started thinking on what would be the best approach, should i run a streamprocessor with FaceSearch directly on the stream, will this be the whole day running?? or can i run the same processor but with ConnectedHome and just look for faces, then when a face is detected run the recognition process and compare it with the ones on a collection? what would be the most cost-effective solution? is there any other option?

regards,

Giovanni Lauria · Answer

For a cost-effective live stream face detection process, use **Amazon Rekognition ConnectedHome** to detect faces continuously, then trigger face recognition against your collection only when faces are found. This avoids running the full recognition process all day, reducing costs. Alternatively, you could schedule or use event-driven processing to further minimize resource usage.

Hiram Acebo · Answer

Thanks!, I am doing a test with ConnectedHome but i need an starting and stopping point, would that mean that I need to schedule the process to run with a cron job or something? or once started will run the whole day until i call stopprocessor?

Run a face detection process on live stream

Relevant content