Ivideon or other cloud based recorders

[marley0ne]

Hey everyone I get a lot of residential requests for systems like Dropcam or Arlo as they want better email reporting and alerts. Normally use Hikvision but you tend to get many false positives with motion detection.

A client I do work for is requesting something to similar to the Arlo at his house. I started looking into things and see Ivideon makes cloud management for Hikvision camera. It looks like this has smarter motion and alerting. Anyone try our similar?

 

[nayr]

motion detection is always going to be susceptible to too many false alerts..

Use cameras that support TripWire/Intrusion detection and you will have better results.. or deploy BluIris servers.

 

[fenderman]

Arlo or anything similar is a terrible idea.
1) Motion detection only works when close to the camera, because it relies on PIR only. You also dont get any pretrigger frames.
2) no continuous record.
3) MONTHLY FEE!!! If you want more than 1gb or 7 days of recording you have to pay monthly.
4) Battery replacement, time, cost, Murphy's law-they will die when you need it
5) Cannot display 24/7 on internal monitor
6)Cannot integrate into any other system.
7) wifi is unreliable
8) Its way over priced, it would cost about the same to pay someone to run ethernet.
Blue iris can get very accurate motion alerts if setup correctly or use PIR motion sensor with it......You might also look at sighthound which claims to have human detection analytics...

 

[riceandbeans]

I have sighthound deployed within a VM and have experience with it. Vastly superior detection analytics to blueiris (supports object classifiers, people only, currently), but also vastly inferior 'other' features (scripting, cam management, etc). I use the two in parallel to get the best of both worlds.Detection of a person is very much like cancer detection. It's a skewed binary classification problem; you're finding the needle in the haystack. If you create a system with no false positives, chances are you are also missing true positives as well (this is referred to as your 'precision'; TP / (TP + FP))You can stack the odds in your favor with good lighting, camera positions, and settings in BI. Properly set up, you can have an extremely sensitive system that also returns very few false positives. Since, similar to cancer detection, the consequences of a missed TP are much worse than a returned FP, you want to be conservative with your settings to be slightly more 'tightly wound'. Any TP alerts I keep in an archive as testing data. Then, if I make changes to my detection settings, I will always use this reserved testing data archive to ensure my new setting detect historic TP conditions. Some other commercial packages support retroactive training of filters, however BI does not. GL and HTH