[tool] [tutorial] Free AI Person Detection for Blue Iris

Joined
Oct 17, 2020
Messages
1
Reaction score
0
Location
Kuna Idaho
Thank you for doing this its awesome. I have one problem though. My bot keeps sending me this message
r2d2, [17.10.20 21:48]
[17.10.2020, 21:48:09.037]: WARNING: No default camera found. Aborting.

He is getting to be quit annoying can anyone help with this.
 

Village Guy

Pulling my weight
Joined
May 6, 2020
Messages
291
Reaction score
161
Location
UK
@Village Guy Nope. no change when I reset Iframe?? I did notice however in BI if I go to the Video tab, select the force size box, change it to match the mask size, which of course is the same size as the full res pic and click OK, then go back in it is not staying at the resolution I changed it to, it defaults back to 768X432, does this on all the cameras. That is also the size of the (sub-stream?) pics it is picking up and throwing in the AI folder. Do you , or anyone think this may be a BI issue? Easy fix is to stop running the sub-stream but it is jacking w/ head now and I want to know WTH is going on. It is my understanding it should only be putting the main stream photos onto the AI folder am I correct?

Unrelated to this annoyance I am still getting this- I have seen some others asking about it but I have not (sorry if I missed it) a reason of why it throws the error

Starting analysis of D:\AIinput\LSyardsd.20201017_191511557.jpg
[17.10.2020, 19:15:12.033]: System.IO.IOException | The process cannot access the file 'D:\AIinput\LSyardsd.20201017_191511557.jpg' because it is being used by another process. (code: -2147024864 )
[17.10.2020, 19:15:12.041]: Could not access file - will retry after 10 ms delay
[17.10.2020, 19:15:12.074]: Retrying image processing - retry 1
[17.10.2020, 19:15:12.079]: System.IO.IOException | The process cannot access the file 'D:\AIinput\LSyardsd.20201017_191511557.jpg' because it is being used by another process. (code: -2147024864 )
[17.10.2020, 19:15:12.083]: Could not access file - will retry after 20 ms delay
[17.10.2020, 19:15:12.120]: Retrying image processing - retry 2
[17.10.2020, 19:15:12.129]: (1/6) Uploading image to DeepQuestAI Server
[17.10.2020, 19:15:12.949]: (2/6) Waiting for results
[17.10.2020, 19:15:12.961]: (3/6) Processing results:
[17.10.2020, 19:15:12.967]: Detected objects:
[17.10.2020, 19:15:12.984]: (6/6) Camera LSyardsd caused a false alert, nothing detected.
[17.10.2020, 19:15:12.992]: Adding false to history list.
To the best of my knowledge the retries are not errors as such. I suspect the file might not have being writen to disk or it is being indexed by the OS after being recently added which might make it unavailable. I can confirm that I see the same retries in my log.

Can I propose that you switch cameras with one channel that is working, this will rule out a configuration issue. Just temporarily swap the ip addresses of a good and bad camera see if the problem follows the address or the camera configuration.

How many cameras do you have that work versus those that error? As I recall you mentioned earlier that only one channel is having an issue.
 
Last edited:

balucanb

Getting the hang of it
Joined
Sep 19, 2020
Messages
146
Reaction score
23
Location
TX
To the best of my knowledge the retries are not errors as such. I suspect the file might not have being writen to disk or it is being indexed by the OS after being recently added which might make it unavailable. I can confirm that I see the same retries in my log.

Can I propose that you switch cameras with one channel that is working, this will rule out a configuration issue. Just temporarily swap the ip addresses of a good and bad camera see if the problem follows the address or the camera configuration.

How many cameras do you have that work versus those that error? As I recall you mentioned earlier that only one channel is having an issue.
6 cameras total, I think over time I have seen this particular issue on most of them at one time or another. I will need to scrub the logs and really pin down if all or just some are showing this and try your suggestion, possibly just overloading the system with too many things at once? It does appear to keep trying and finally upload it to DeepQuest...
 

jj2473

n3wb
Joined
Dec 30, 2019
Messages
8
Reaction score
0
Location
New Jersey
Hello Everyone,
Using Aitool, upload to Telegram works fine. If I close out and use the Vorlon fork with the same Telegram token etc. in the Vorlon fork I get the following:

DType Time Idx Detail MemberName
Error 10/18/2020 9:55:30 AM 3 ERROR: Could not upload image C:\BlueIris\aiinput\AIGarageCam1.20201018_090120905.jpg to Telegram: chat not found [ChatNotFoundException] Mod: <TelegramUpload>d__32 Line:1460:33 TelegramUpload>

Any thoughts? I have copied and pasted to rule out typo's, and manually entered it. Always fails under Vorlon but works under original ai tool.

Thanks in advance.
 

Village Guy

Pulling my weight
Joined
May 6, 2020
Messages
291
Reaction score
161
Location
UK
6 cameras total, I think over time I have seen this particular issue on most of them at one time or another. I will need to scrub the logs and really pin down if all or just some are showing this and try your suggestion, possibly just overloading the system with too many things at once? It does appear to keep trying and finally upload it to DeepQuest...
Why not set the read ahead on all cameras to 0 (switch it off) except the most active. If it works reliably, add say two more cameras and so on until it breaks or perhaps works just fine! I'm wondering if BI is having difficulty due perhaps the cpu running out of steam?
 

balucanb

Getting the hang of it
Joined
Sep 19, 2020
Messages
146
Reaction score
23
Location
TX
Why not set the read ahead on all cameras to 0 (switch it off) except the most active. If it works reliably, add say two more cameras and so on until it breaks or perhaps works just fine! I'm wondering if BI is having difficulty due perhaps the cpu running out of steam?
I may try that, thanks. I'm running a i7 with 32mb of ram and the only thing on that computer is BI/AI/DS you would think that would be enough juice, perhaps not.
 

Village Guy

Pulling my weight
Joined
May 6, 2020
Messages
291
Reaction score
161
Location
UK
I may try that, thanks. I'm running a i7 with 32mb of ram and the only thing on that computer is BI/AI/DS you would think that would be enough juice, perhaps not.
I agree it should be more than enough. Just two last thoughts, are you running the latest BI Version and does the file size change. When I had issues earlier on it turned out to be a BI problem. When I looked at the files captured for deepstack and it clearly showed up the issue as occasionally a camera file would be much smaller than expected.
How else could you have so much fun:)
 

balucanb

Getting the hang of it
Joined
Sep 19, 2020
Messages
146
Reaction score
23
Location
TX
I agree it should be more than enough. Just two last thoughts, are you running the latest BI Version and does the file size change. When I had issues earlier on it turned out to be a BI problem. When I looked at the files captured for deepstack and it clearly showed up the issue as occasionally a camera file would be much smaller than expected.
How else could you have so much fun:)
I was talking about 2 different things (Problems) shouldn't have done that since I am replying to both and if anyone else is trying to follow, it will confuse them! Sorry everybody. :) On the mask vs. the AI image problem- - Running the newest version of BI, have not checked the file size, the image size is smaller however (maybe that is what you mean?) and yes that is the "cause" of the error, BI is dropping the sub stream pic in the AI folder, AI looks at the image and says HEY! this pic and your mask size don't match!!! (Danger Danger Will Robinson - Hope your old enough to get that ref. LOL )- This should not happen as far a I understand it should only be putting the HD pics in there!?. I really think the root cause of this is coming from BI and not the other things (AI/DQ) I sent a Help/Bug request to BI... I decided to just remove the sub stream from the 2 cameras giving me fits, so yeah I gave up, but that "fixed" it. On the System.IO.IOException....because it is being used by another process. (code: -2147024864 ) error Again not really sure what is going on, wish I was smart enough to figure out what the "other process" was, the image is eventually getting processed so not so much a error as annoying, to me anyway. I'm putting both of these to rest for now. I just switched back (this morning) to the VolronCD version of AI, lets see what issues we run into now :eek:. Unfortunately I fall deeply into the operator category on all this stuff, no clue how all the code operates, REALLY appreciate the patience and help on all of it though!!!
 

Tinbum

Pulling my weight
Joined
Sep 5, 2017
Messages
446
Reaction score
126
Location
UK
I just switched back (this morning) to the VolronCD version of AI, lets see what issues we run into now :eek:.
Just remember that if your not running the very latest, with all its commits, then the issue may have already been sorted,
 

aesterling

Getting comfortable
Joined
Oct 9, 2017
Messages
352
Reaction score
346
Deepstack's development has seemed stalled since earlier this year and it was starting to look abandoned, but I tagged the two developers over on Github and one of them (John) replied with this comment:
Hello @AndrewHoover @aesterling @VorlonCD

Thanks for bringing this up and it is exciting to see how DeepStack is being used.
We truly haven't done much development as we would have wanted to in recent times. It took a bit of settling down for us into our new job. Apologies for the pause in recent times.

We are planning on stabilizing the development and would have new releases in coming weeks with significant improvements. The project is not abandoned and a lot is coming on it soon.

Please bear with us.
Thanks
Obviously, there are other engines available and more to come, but many of us are actively using Deepstack so it's good to hear they'll continue working on it.
 
Last edited:

Tinbum

Pulling my weight
Joined
Sep 5, 2017
Messages
446
Reaction score
126
Location
UK
tried building in visual studio the current release plus all commits but failed miserably :(

Please could someone spare the time to knock up an idiots guide to building this from git hub with the latest's commits, I and I know a few others will be very gratefull :)

edit or the real easy way :) :)
 
Last edited:
Joined
Oct 1, 2020
Messages
18
Reaction score
2
Location
Texas
So I've been messing with other options that could utilize my GPU on a Windows Server that doesn't have AVX... Sadly this mix doesn't work at all with DeepStack currently and considering where the software is at this point I've been looking at TensorFlow.

How much of an issue would it be to add an option for a TensorFlow server, as it seems it works very similar to DeepStack where you can feed in an image and it'll spit out what it found. I think giving people a second option, as long as it isn't too hard to implement, on what server they use would be very helpful. TensorFlow could be better for some people too as you can build your own models.

Edit: One reason I bring this up is over the last week I have been playing with using TensorFlow and Blue Iris. In every case the examples I already found online are over a year old and A LOT has changed between TensorFlow and Python, in every case something is broken and I can not fix it. In many cases I manage to get the TensorFlow parts working but something else crashes. Getting a TensorFlow server up and running seems like the easy part compared to feeding it images, getting results and integrating it into Blue Iris.
 
Last edited:

Eatoff

n3wb
Joined
Aug 28, 2020
Messages
19
Reaction score
3
Location
Australia
This is intentional. When I asked Ken to implement the timeline clearing I actually requested he leave the alert list images because otherwise you now have no way to know about false negatives occurring. Leaving them allows one to see that an alert was canceled that maybe shouldn't have been and thus you can take action. This I felt was a good compromise because the timeline is now cleaner and you can jump clicking on that. If he removed the images on cancel you'd never know if it was canceling alerts it shouldn't have potentially. Keep min mind AI Tools isn't the only applicable use here so you have to think of it as a best compatibility overall vs. design thing because another use case may not use AI Tools and thus you don't have that to ref back for those said false negatives.

If you don't want to see them you can still use the flag=1 to flag and just filter that image list to use flags only and you won't see them.
I see what you mean, but the webUI no longer lets you see just the flags. Its alerts or clips, so all the false triggers are there. Same goes for the BI app (android version anyway). So i see a whole lot of alerts, and not one of them is actually a positive.
 

105437

BIT Beta Team
Joined
Jun 8, 2015
Messages
1,995
Reaction score
881

edit or the real easy way :) :)
Just installed this version. Launch AI Tools and it's stuck "Initilalizing Database" in the History tab.
 

balucanb

Getting the hang of it
Joined
Sep 19, 2020
Messages
146
Reaction score
23
Location
TX
Can we talk about what right looks like- I am getting alert images sent to Telegram and as far as I can tell they are indeed relevant and I don't have issues with that. I am (now) using the VolronCD version. On the Stats page you have the circle chart and on the bottom it shows # of errors, Alerts, Irrelevant, false, skipped img queued and Action Queued.
Here are my questions-
1. I am getting a ton of errors all showing the trigger URl's, I know the data is correct, I am using both of the URL's from the original set up from pg 1 of this thread, with this version should I only be using 1 of them or should they be different than they are on the original instructions? Should I be putting something in the Cancel URL block (on the Actions Screen)
2. I get thousands of irrelevant and false alerts- they don't end up being sent to me as a alert but the high number! Is this normal, Obviously I don't want to get alerts on them but the #'s seem so high I wonder if I am doing something wrong.
3. What are the skipped files about? Why are they skipped-Should be concerned?
4. IMG Queued/Action Queued? What is their purpose?

As always TIA!!
 
Top