Hi all! Just purchased the BI.

[Pentagano]

HI all,

Just into my 5th day of trying out BI. So far very happy.

Joined this forum and saw the discount so just purchased and registered the product this morning.

Exhausted all other options - motioneye/ipsy/motion/zoneminder/frigate etc.
I use homeassistant also so have mqtt integrated with nodered etc.

Having a cheap and cheerful dahua nvr for constant recording of my cctv I wanted much better motion and object detection.

Now I have BI running as a service on a i7 7700 32GB dell. Homeassistant is on a mini pc 4GB with small ssd (this was not suitable for any image processing).

I have used the fork of the AI tool for processing (jpeg snapshots taken every 2 seconds upon any motion trigger). Also deepstack as a container on my homeassistant mini pc with ubuntu.
AI tool is running as a service also.
Just restarted my dell and all services started well. Can assess via web also. (using nginx proxy/dns name on the ubuntu so no open ports on the gateway).



Using AI but I'm finding detection is not very good at nighttime so far. BI standard motion detection is good at night time but too sensitive during the daytime.
I have created now different profiles for night/day.


Trying to detect this cat using the AI tool but comes up with false alert even though the night time setting of the BI detects it (black and white with settings at max sensitivity).
Are there any settings I have missed in the AI tool (vorloncd) that may help it capture this animal at night time? I'm assuming it is just too dark and far away for the AI tool?




Thanks

 

[Holbs]

soon to be getting into HA/MQTT/NR (have dabbled but nothing serious). Can toss up some of your NR flow snapshots?
Sorry, do not know anything about AI as of yet.

 

[IAmATeaf]

I found similar issues using AITools, was perfect during the day but at night it would miss things.

For me and most likely the same for you it’s to do with the contrast of the picture where DQ just can’t differentiate an object.

I moved over to using IVS on the cams using tripwires which not 100% has minimised some of the false alarms for me. Headlights in the evening is still an issue but haven’t really had the time to fine tune things.

 

[Pentagano]

I found similar issues using AITools, was perfect during the day but at night it would miss things.

For me and most likely the same for you it’s to do with the contrast of the picture where DQ just can’t differentiate an object.

I moved over to using IVS on the cams using tripwires which not 100% has minimised some of the false alarms for me. Headlights in the evening is still an issue but haven’t really had the time to fine tune things.

Going to install a low wattage led lamp in that area tonight and see how it goes.

Can you elaborate on the ' IVS on the cams using tripwires' please? What is this?

Thanks

 

[Pentagano]

Possibly a way of altering the contrast also of the jpeg snapshot taken by BI before processing?

 

[Pentagano]

What do these functions do? I see options for brightness and contrast in the drop down:

 

[Pentagano]

When the profile is greyed out does this mean 'selected' or the reverse? I'm assuming greyed out is disabled

 

[Pentagano]

hmm tried enabling the PTZ controls and maxing the contrast/gain - no change

Most likely my cameras do not support this but thought the blue iris could sharpen the images for processing?