Meraki Vision object detection

cmr
Kind of a big deal
Kind of a big deal

Meraki Vision object detection

How successful have you found the person and vehicle object detection to be in the Meraki Vision portal?

 

I would say the person detection is quite good, though the colour options on the 3rd generation MV73X camera is a little off with bags and other carried items counting as clothing.   However, the vehicle detection is pretty poor with, in my case cats being classed as cars and my wife being a motorbike!  This is all in a clear daylight shot that has no other 'noise' as it is a private courtyard space.

 

What's your experience? 

If my answer solves your problem please click Accept as Solution so others can benefit from it.
9 Replies 9
alemabrahao
Kind of a big deal
Kind of a big deal

I think this is a bit tricky, as accuracy depends heavily on camera angle, lighting, and movement.

What you can try is adjusting the camera angle to get a clearer side or top-down view of vehicles, and try using zones of interest to limit detection to specific areas.

I am not a Cisco Meraki employee. My suggestions are based on documentation of Meraki best practices and day-to-day experience.

Please, if this post was useful, leave your kudos and mark it as solved.
cmr
Kind of a big deal
Kind of a big deal

@alemabrahao I guess it might help, but unless a developer made a typo, this trace is definitely a cat and not a car...

 

cmr_0-1751058932127.png

 

In fact, if I use the rough match to exact match slider, at rough match over three days I get 40 matches which includes all the cars and some cats.  As I slide towards exact match the cars disappear and as I approach exact match only the cats are left...  I think a developer might have had a laugh here! 

If my answer solves your problem please click Accept as Solution so others can benefit from it.
cmr
Kind of a big deal
Kind of a big deal

Scratch the exact match issue, that only occurred when I searched on a cat 🙈, so not surprising that it matched other cats!  Still unsure as to why they appear in the first place... 🤔

If my answer solves your problem please click Accept as Solution so others can benefit from it.
alemabrahao
Kind of a big deal
Kind of a big deal

I think it's totally possible that it's a car, check this out. 🤣🤣

1000114262.jpg

I am not a Cisco Meraki employee. My suggestions are based on documentation of Meraki best practices and day-to-day experience.

Please, if this post was useful, leave your kudos and mark it as solved.
RWelch
Kind of a big deal
Kind of a big deal

😂 🤣 😂

If you found this post helpful, please give it Kudos. If my answer solves your problem please click Accept as Solution so others can benefit from it.
Dunky
Head in the Cloud

hahaha, I get pigeons, magpies, badgers, foxes and muntjack deer triggering alerts when its set for person detection only.  Fortunately these are not too frequent.

PhilipDAth
Kind of a big deal
Kind of a big deal

>vehicle detection is pretty poor 

 

I've mainly used the cameras for number plate recognition (and this was several years ago).  The camera needed to be as low as possible to make it as inline as possible with the car.

 

The cat recognition is pretty funny.

 

DevOps
New here

I've been piloting a single camera at our staff entrance in our HQ, and during the day, the people detection and analytics (In/Out) has been pretty solid. However, during the night on occasion we have it saying there are people entering and leaving, even though when you search for People it shows none. It as if the 'crossing the line' is being triggered by an imperceivable to  the human eye change in brightness, but the programming isn't validating the count by checking that there is a person in view. I have a call open with Meraki, but unfortunately apart from it being raised up the chain, not much response back.

PhilipDAth
Kind of a big deal
Kind of a big deal

It definately needs lighting (or IR lighting) to work nicely.  🙂

Get notified when there are additional replies to this discussion.
Welcome to the Meraki Community!
To start contributing, simply sign in with your Cisco ID. If you don't yet have a Cisco ID, you can sign up.
Labels