Search results

Unfortunately this wasn't it. I hope it's just some setting I can disable. I don't want to be left behind in updates. Seems no one else is having this problem. It disappears if I disable ai, so it's something related to that, I believe.

what happened between blue iris 5.8.6.x and 5.8.7.x that could cause a massive increase in my codesense AI response time? only new thing i saw in the AI panel was pre-trigger images, but even with that set at 0 i went from 32ms average to 800ms+. I'm also getting CPU spikes 80%-100% when the...

Single

Anyone have a massive memory leak issue with cpai/coral on unraid? After about 18hr my unraid rig becomes unresponsive and I see that the cpai docker is taking up 10gb+

Is it possible to run instances of cpai on different servers and specify which IP blue iris uses per module? For example face detect on my windows PC with a GPU, and everything else on my Linux box with the coral tpu

Same issue here. Once I train faces, it no longer adds new faces to unknown folder

Thank you. I couldn't get it working at the default 192.0.0.1.

I have the same issue intermittently. It works when I uncheck/check save faces, then randomly stops working until I do it again. It seems to be a conflict between BI storing known faces in its own folder or training faces through the cp.ai dashboard. It's confusing why BI hasn't settled on one...

you've updated this exact laview cam with the latest g1 firmware? i see there are a few versions for "g1" listed as latest a bit confusing link: Firmware | Hikvision USA | The world’s largest video surveillance manufacturer IPC_G1_EN_STD_5.6.6_210625 IPC_G1_EN_STD_5.6.3_190923...

ignorance is bliss, i guess?

You manually edit the modulesettings.json in the appropriate module's folder. Although I doubt that is your issue :(

Yes, it defaults to medium, but I manually set it to small. Medium results in very high latency, worse than even an old p400 video card, so it defeats the purpose. If the coral can't reliably be used at "small," there isn't much purpose for it. The alternative is that the default trained...

just a bit more anecdotal evidence -- After about two weeks or so of constant Coral TPU (small) detection, I have not picked up a single cat or dog. With the Yolo models on CUDA, i used to pick up several per day -- usually accurately. I'm also rarely picking people up walking at night. I...

To clarify, it's 20-30ms on the cp console, and 100ms average on the blue iris status window. Small and tiny are about the same for me, so I'm just using small.

Hey, it's using my substream which is 640x360. I haven't increased the resolution to test if it tanks the performance

A 1060 is about $70, not several hundred

Thanks, but Chris just confirmed the latency we are seeing is expected for medium models

Well, on small I seem to be having accuracy issues. It's very easy for me to test because I'm on a high traffic street. I get a lot of false "bus" and "cat." Not looking too good here ;(

Well that makes sense. He should have said so lol. A few guys are bewildered why the performance is terrible. I assume frigate uses the 300x300 models as well, unless they had a coding breakthrough of some sort

What kind of performance are you guys getting with the coral? It's pretty abysmal (~300ms) on my end unless I set the models to "small." It ships with them defaulting to "medium," so I'm confused about the amazing performance Chris was getting.

Ah how unfortunate. Well, it's definitely progress!

Wait a minute.... THERE'S CORAL SUPPORT NOW? (re: 2.1 changelog) Does it only work on a pi? Well, I assume not because that would be silly. I've had a usb coral sitting in the drawer for months just waiting for this glorious day! Please confirm if true

It's just a gimmick at this point. It works best with a close up cam like a doorbell. Either way, too many false triggers and misses, even with many many photos trained

I process more images for better accuracy. Doorbell is most important to me because it handles the facial recognition. Even with 20 images it still misses sometimes!

Going to have a super long queue if you process a lot of images per trigger. I do 20 images on the doorbell and 10 on the street. Sometimes cameras trigger at the same time within quick succession.

I'm also wondering if there's a "more accurate" setting for facial recognition. I noticed there's a Mode=MEDIUM, but I'm not sure what adjusting that actually does.

I base it on the blue iris AI status page where it has t_ave.

The 5s models brought my response times back to status quo around 350-450ms. Thanks for the suggestion

I'll give this a shot later today. Thanks!

I'm running gpu with an old p400. It may be that sensai is catered toward newer CUDA gpus? My gpu time seems to be on par or worse than others' cpu times, but it is worth the tradeoff of not having my CPU pegged at 90%+ all day. I go through about 1000+ processed custom objects per hour...