We had a question from a potential customer asking if it would be possible with Meraki MV cameras to have auto tracking across multiple camera environments.
The use case would be;
1. Person walks out to kitchen.
2. Person walks out to front door.
3. Person walks around house.
4. Person walks into back door.
So instead of pulling up footage from multiple cameras manually to essentially "follow" them. You would select the object and across multiple camera you could follow them as they progress around the house or camera views, without having to manually select cameras.
Is this achievable or something in pipeline? @GeorgeB
I doubt that. While the cameras do know the concept of people, they cant actually recognize individuals for privacy reasons. For it to follow people it would have to recognize individuals and correlate the imagery of those individuals across multiple cameras.
Hi @Priesty not at this time, this is purposely designed for anonymized person detection, so more for people counting analytics, to be able to tell how many "human movement" events there were on specific cameras at specific times. It is somewhat designed to NOT track people for the various legal/privacy concerns.
But now as MV Sense has been launched, this opens up additional abilities to solve more business problems. MV Sense is a new API for you to interact with and build tools based on the person detection analytics already collected by the cameras. So you can get 3 things: historical info (how many people were here at time X), current snapshot (how many people are here now) and a sub-second MQTT feed of people and location, so you can already get a stream of the number of people and their locations within the frame, pushed in near real time.
The MV Sense use cases I've heard so far are things like triggering a media display when more than 10 people are within range of the screens in a retail environment. Or understand if people are still inside a building that has been evacuated. Or set off an alarm if the number of people near a dangerous piece of equipment falls below a safe threshold. Things of that nature. I do not believe it is possible with the system as it stands today, to actively track specific people across multiple cameras as they proceed through a building for example.
You could put all the cameras in a wall (assuming there aren't too many). Then you could watch all the cameras at once, and as they move from one camera to another.