5.5.8 - June 13, 2022 - Code Project’s SenseAI Version 1 - See V2 here https://ipcamtalk.com/threads/codeproject-ai-version-2-0.68030/

[joshwah]

Exciting times... unfortunately, for me, comparing Deepstack CPU vs SenseAI CPU - SenseAI compared much worse.. I have a Nvidia 1650 SUPER and could never really get the GPU version of Deepstack working.. so keen to try out SenseAI GPU

 

[astroshare]

Just wondering, are the settings below ok for SenseAI ? These are the same I was using with Deepstack and worked great for me, but DS was using an NVIDIA GPU.

 

[MikeLud1]

Just wondering, are the settings below ok for SenseAI ? These are the same I was using with Deepstack and worked great for me, but DS was using an NVIDIA GPU.

View attachment 131004

Settings should be fine for SenseAI. One setting you can change is you can unchecks Use main stream if available. If this is checked it will only slowdown DeepStack or SenseAl because they both will reduce the resolution to 640 x 640 before it does the detection.

 

[astroshare]

Thank you!
Can you disable some of the extra SenseAI features that show on the WebUI, like Background removal, Face Processing, etc? Do they use any CPU?

 

[MikeLud1]

Thank you!
Can you disable some of the extra SenseAI features that show on the WebUI, like Background removal, Face Processing, etc? Do they use any CPU?

I posted the request

 

[MikeLud1]

I posted the request
View attachment 131007

They will be adding the ability to disable modules. It is look like SenseAI is going to be our go to for Blue Iris AI

 

[sebastiantombs]

While it's nice to see another integration that will work with BI I think this was a very premature step on Ken's part. It clearly isn't "ready for prime time" just yet and the integration/switchover in BI isn't "ready for prime time" either.

 

[MikeLud1]

While it's nice to see another integration that will work with BI I think this was a very premature step on Ken's part. It clearly isn't "ready for prime time" just yet and the integration/switchover in BI isn't "ready for prime time" either.

Agree, I think soon it will be ready for prime time. The developer is very open for suggestions and is implementing them. Since the developer is currently doing updates I think this is the time to request our suggestions.

 

[wittaj]

Could he incorporate your LPR stuff into just a check box like what exists for Plate Recognizer?

You check that box and behind the scenes it makes all the stuff it takes to create your system?

We have to remember that DeepStack when it first came to BI wasn't "ready for prime time either" and it got better faster with each revision.

 

[sebastiantombs]

Yes, that's true of DS as well, but it did work on the first release. This time is different because DS already has been integrated with BI and no provision was made in BI to "decouple" the two when senseAI is installed. Not a good thing, especially when there didn't seem to be much in the way of instructions in the update notes for BI regarding that.

 

[MikeLud1]

Could he incorporate your LPR stuff into just a check box like what exists for Plate Recognizer?

You check that box and behind the scenes it makes all the stuff it takes to create your system?

We have to remember that DeepStack when it first came to BI wasn't "ready for prime time either" and it got better faster with each revision.

I am going to suggest adding an ALPR. I am going to share my DeepStackALPR code with them. I think it will not be that difficult for them to implement. SenseAI is Open Source so I may give it a try to make a module, but my programing skills are not the greatest.

 

[wittaj]

I am going to suggest adding an ALPR. I am going to share my DeepStackALPR code with them. I think it will not be that difficult for them to implement. SenseAI is Open Source so I may give it a try to make a module, but my programing skills are not the greatest.

I know many here are very appreciative of the efforts you have done on this and works well for many people, but if they could implement a check box type solution, that would get this into so many more hands and systems!

 

[jrbeddow]

Ok @wittaj , we get it, with all due respect that's about the fourth or fifth time you have suggested the checkbox for LPR on the forum. Wait a bit, then if it doesn't happen soon enough, send a formal request to Ken and/or the CodeProject SenseAI developers.

 

[wittaj]

Ok @wittaj , we get it, with all due respect that's about the fourth or fifth time you have suggested the checkbox for LPR on the forum. Wait a bit, then if it doesn't happen soon enough, send a formal request to Ken and/or the CodeProject SenseAI developers.

Wow, sorry a few mentions offended you LOL. But I see you are from Cali, so that explains being easily offended

Over the course of several months, I mentioned it like twice regarding when it was DeepStack and twice looking at it as Sense AI - one was to their developer and one was to @MikeLud1

And how do you know I haven't made a formal request to the developers? Sometimes the only way things get added is if enough people request it. So my posts may put the idea into others to formally request it as well...

It is apparent at the moment this thread is getting action on SenseAI so I put in the request and reminder to see if something can happen.

 

[Tinman]

I'm not blaming Ken at all on this. He stated that if you were relying heavily on the custom AI models that you may need to stick with Deepstack. I did the upgrade to 5.5.8.0 and am still running Deepstack just as I was on 5.5.7.11. All he did for now was make it possible for those who like to test the ability to try Code P. I'm just glad Blueiris keeps moving forward

 

[jrbeddow]

Agreed, I'm not blaming Ken at all, and am happy to see choices moving forward. The current implementation is a bit rough in the sense that it makes it somewhat hard to move back to DeepStack (as it forces a complete uninstallation of AISense), but that will likely improve with time. We have to remember this new option only started this week on Monday, so it has only been about 4-5 days.

 

[lane777smith]

Below is how I have mine set, make sure you have default object detection checked

View attachment 130635

Change 127.0.0.1 to localhost and uncheck Auto start/stop option. That got SenseAI working for me BI instance.

 

[joshwah]

We also need to remember, this new integration is not available on a “stable” version of BI… it’s on the bleeding edge / beta versions essentially.

 

[MikeLud1]

The next release of SenseAI will have custom model support. I am not sure about GPU support, it may.

 

[tofu]

No rush... Doesn't seem to do anything better than deepstack at this moment. Just something new to tinker with