Blue Iris vs Protect vs Camect

Chapin

Getting comfortable
Joined
Oct 5, 2017
Messages
122
Reaction score
21
I've been a happy BI user for just over a year, big thanks due to this community for helping me get off the ground.

For the past 2 - 3 weeks I've been testing the Unifi Protect system and also the crowd funded Camect NVR. All three of these are good systems and each has strengths and weaknesses that the other don't. I thought I'd share some observations here of these three systems. I'll try to add to these lists as I think of more.

My cameras are Dahua, Unifi, and a Nest Doorbell

Blu Iris Pros:
Most powerful and configurable of the three
Good AI (Sentry); only one with boxes around movement which is really nice

Blu Iris Cons:
No integration of IOT (Nest etc) cams
Most AI false positives
Need to open port or set up VPN for remote access

Camect Pros:
Best camera choices, ONVIF, RTSP, IOT (it's really nice having the Nest appear with the other cams)
Best AI, able to differentiate between a squirrel, cat, and dog; Also apparently can differentiate between many other categories and alert/not alert based on object type.
Best AI notifications
Set up was pretty straight forward. Found all of the cameras, except the Ubiquity which streams on a non-standard port. It found all of the Dahuas which were are on a sequestered vlan, different than the Camect vlan.
Camect support is really responsive to solving issues
Most trainable AI

Camect Cons:
This is an early product, a lot of things could work better or be more intuitive. I think that this will improve over time.
No mobile apps, uses a mobile browser window. I would guess that apps will come along.

Protect Pros:
Simplest Set up
Most intuitive interface (web and IOS)
Best remote viewing experience (iOS), including playing back motion events and scrubbing playback video
Fewest AI false positives

Protect Cons:
Only Unifi Cams
Unifi Cam (UVC G3 Pro) picture quality not on par with Dahua (IPC-HDW5231R-ZE)
 

M-Cube

n3wb
Joined
Nov 29, 2018
Messages
14
Reaction score
6
Location
@ my laptop
Would be nice to have the ability to install the Camect software on your own hardware. For me that's the major Con of Camect at the moment, bound to the limited capacity of the provided hardware only being capable of handling about 24Mp total resolution.
 

Iemand91

Getting the hang of it
Joined
Aug 12, 2016
Messages
146
Reaction score
40
Location
Netherlands
Camect looks really interesting. But I always thought smart motion detection methods (from tripwire to people recognition etc.) took a lot of CPU/RAM power for a 3rd-party NVR.
The benefit of using a same brand NVR (for example Dahua camera's with Dahua NVR) was that the camera's could do the smarter ways of motion detection (IVS) themselves and they could "talk" to the NVR thus reducing the needed (powerfull) hardware in the NVR.

That's why you see the quite powerfull (needed) systems needed for (example) Blue Iris.
Than I'm wondering how this little box with just a Intel Celeron J3455 CPU and 4GB RAM is capable of doing all that.
 

Chapin

Getting comfortable
Joined
Oct 5, 2017
Messages
122
Reaction score
21
Would be nice to have the ability to install the Camect software on your own hardware. For me that's the major Con of Camect at the moment, bound to the limited capacity of the provided hardware only being capable of handling about 24Mp total resolution.
I would guess that future versions of Camect will have more capacity.

I'm pleasantly surprised at how well it works. They have a ways to come, but compared to other tech crowdfunded projects I've supported, this one is pretty good. they do have the best motion detect AI I've used. It differentiates between a cat, squirrel, raccoon, and dog, as well as correctly identifying a USPS truck. It's not 100% yet, there is a cat that has been a dog and a squirrel. but, pretty impressive overall.
 

Chapin

Getting comfortable
Joined
Oct 5, 2017
Messages
122
Reaction score
21
Camect looks really interesting. But I always thought smart motion detection methods (from tripwire to people recognition etc.) took a lot of CPU/RAM power for a 3rd-party NVR.
The benefit of using a same brand NVR (for example Dahua camera's with Dahua NVR) was that the camera's could do the smarter ways of motion detection (IVS) themselves and they could "talk" to the NVR thus reducing the needed (powerfull) hardware in the NVR.

That's why you see the quite powerfull (needed) systems needed for (example) Blue Iris.
Than I'm wondering how this little box with just a Intel Celeron J3455 CPU and 4GB RAM is capable of doing all that.
I used a Dahua NVR before BI. It did not work well for me. sluggish and counter intuitive. BI works much better for me.

I think that all three of these, BI Sentry, Protect, and Camect use the cloud for the AI motion alert processing and not the local NVR.
 

Iemand91

Getting the hang of it
Joined
Aug 12, 2016
Messages
146
Reaction score
40
Location
Netherlands
I used a Dahua NVR before BI. It did not work well for me. sluggish and counter intuitive. BI works much better for me.
I have a Dahua NVR for a while now, but I don't have much experience using it (or maybe experiencing sluggish behaviour), since I usually just use the IVS recordings on the camera's SD-card for watching footage back and that works fine.
I think that all three of these, BI Sentry, Protect, and Camect use the cloud for the AI motion alert processing and not the local NVR.
Not according to them:
The AI analytics are mostly on on the CPU at the moment. We do use the integrated intel GPU extensively for encoding and decoding though. No support for an external GPU yet.
Even more:
Re: "And the camect connects to the internet no matter what?" ....

I might not understand your question, but Camect does require internet connectivity when you first activate it. After activation, it will run fine without connectivity (and will record when your internet is down if you're using it with regular IP cameras) but it does require occasional connectivity in order to download software updates and to do some maintenance things like make sure the time is correct. We're not recommending that people run it in the complete absence of connectivity because we don't test that use case.
I'm not all into the alternative CCTV recording software, but I'm pleasantly surprised that there's software that can handle like 12 2MP camera's on a system with just a Intel Celeron J3455 CPU while doing all that cool smart alerts and even see the difference between animals and stuff.
Usually it takes quite a bit of CPU power to do that, so that is a nice surprise.

It's not for me (yet) since I don't have a big install, but it's definitely a nice development.
 

aristobrat

IPCT Contributor
Joined
Dec 5, 2016
Messages
2,679
Reaction score
2,580
I’ve got a 12-camera system. One 8MP, three 4MP, one 3MP and the rest 2MP. BI shows they generate ~550 MP/s.

My BI PC is an i7 from 2012 (not what I’d consider powerful vs today’s systems) and the processor runs at about 25%, so hearing that a NUC running a current generation Celeron can keep up doesn’t sound all that crazy.

There’s a project someone is hosting on here that uses an open-source AI package you can install on your BI machine that will scan BI alert snapshots. It’s not directly integrated with BI so there’s a fair amount of manual setup required initially... IMO that’s where off-the-shelf solutions like Camect are going to be nice for the masses that aren’t interested in tinkering with stuff like that.
 

spectator

n3wb
Joined
Oct 19, 2017
Messages
23
Reaction score
17
I think that all three of these, BI Sentry, Protect, and Camect use the cloud for the AI motion alert processing and not the local NVR.
Not Camect. Your video is never sent to the cloud. Camect AI is 100% local to that little box with a Celeron J3455 and 4GB of RAM :). It will work even if your internet connection is down. Without internet, you won't be able to get notifications or email alerts, but you can review the alerts in the UI.

This is why, unlike other NVRs, CPU is the limiting factor for Camect ... The Indiegogo unit has a recommended usage limit of 24MP of total camera resolution, assuming "normal home security" usage (i.e. it could be lower if you have cameras looking out on scenes that are constantly busy and require constant processing by the AI).
 

Oceanic

n3wb
Joined
Nov 4, 2019
Messages
1
Reaction score
0
Location
London
Hi Chapin,

I am waiting for my Camect to be delivered, where you able to connect any Ubiquity camaras to Camect?

Thanks for posting your experience on the three systems, very useful
 
Top