I would have thought the dynamic mask will mask out the car when stationary ( after a couple of hits or whatever you set) but when it moves it will record as its now outside the mask. I don't think the person and 5 minutes should be involved.
[tool] [tutorial] Free AI Person Detection for Blue Iris
- Thread starter GentlePumpkin
- Start date
I'm running the CPU version of Deepstack on my Windows 10 PC with BI. I just enter all three locations in AITool and it load balances automatically... probably just round-robin and not actually looking at resource indicators.
iankaufmann
n3wb
I am experiencing this at the moment... I know it's an older post.
Clicking "test" works perfectly, but when the time comes for a real detection, nothing is sent out over MQTT... not sure what I'm doing wrong.
I have been using AiTool since 2.0.720 and MQTTT messages have worked for me with no issues. Make sure that on the Camera Actions tab you configure the triggering objects for MQTT and set the object priority and threshold the same or lower than under Relevant Objects. For example on the Cameras tab under Relevant Objects if you have the minimum threshold for person set at 40% but on the MQTT Triggering Objects the threshold of a person is say 60% you have a gap between 40% and 60% where you would not get any MQTT topics and payloads for person. Make sure your logging level is set to debug. You might disable Dynamic Masking for this test. A difference in the log for the Test and an image that triggered an alert will give you a place to start.
iankaufmann
n3wb
It seems like maybe it has to do with the fact that I tried to change the MQTT topic (just out of the irrational desire to standardize some syntax across everything I do).
Reverting things back to ai/[Camera]/motion seems to have fixed this problem. Still strange that it worked when I was hitting "Test"... but there must be something internal somewhere on the "real" notification. It also was still executing the trigger URL and everything on the BlueIris side was working fine... just nothing going to the MQTT broker.
Anyway, it appears to be working now with the default MQTT topic! Now I can have a binary sensor in Home Assistant specifically for vehicle and person in the driveway.
Thanks for the input.
I am triggering BI via HomeSeer. For example if the camera detects activity in the yard in the daytime when we are home BI is not triggered unless a car has been identified within the last few minutes (car pulling into the driveway and someone getting out) then I trigger the doorbell chime. HomeSeer triggers BI whenever a car pulls into the yard but I am not interested in recording the family in the yard. Another example is that I only want to record animals in the yard when we are not home or at night when we are sleeping. You can really cut down on the number of alerts that BI generates if you trigger BI with additional conditions in HA.
iankaufmann
n3wb
Yeah that's basically what I'm up to here too!
As far as notifications on mobile, AI Tool was an absolute game-changer. I don't even remember the last time we got notifications on our phones that weren't wanted. Prior to AI Tool, I was constantly battling the detection settings in BI between not high enough (missing obvious alerts) and being too sensitive (getting notified for every gust of wind and cloud change).
With AI Tool flagging alerts, and notifications only going out for flagged alerts, the problem is completely solved.
However, the Blue Iris integration for Home Assistant doesn't give me the specific detection from AI Tools that caused the trigger, and I wanted to play different sounds on the indoor door chime when a vehicle enters the driveway, or a person is approaching the doorbell. (Just a subtle notification sound in case we don't have phones in-hand)
So, being able to have AI Tool call out directly via MQTT is what did the trick there.
sunnyhaibin
n3wb
- Jul 12, 2017
- 12
- 0
After running VorlonCD AITool for about a week, RAM usage stays super high and does not go down, I will have to close out AITool every so often to release the RAM. It has been doing that on 2 different machines after a short period of usage.
Does anyone have this issue with the latest version of VorlonCD AITool? What can I do to allow it releasing the RAM? Thanks in advance!
Does anyone have this issue with the latest version of VorlonCD AITool? What can I do to allow it releasing the RAM? Thanks in advance!
I triggered 4 cameras twice. They are triggering every 2 seconds. Below are the resources I am using. AiTool goes from 165.7MB at idle to 167.8MB under load. Did you get your DeepStack and Docker issue resolved.
Just looking through my history tab in AItool now to check the triggers etc. but every time there is a new motion alert it jumps all the way back to the bottom, which makes it hard to find specific events. is there a way to stop this happening?
build 2.0.943.7743
build 2.0.943.7743
- Sep 4, 2017
- 194
- 325
I'm sorry you seldomly hear things from my end and thankfull VorlonCD is doing such a great job! Today, I got myself a Jetson Nano running too and I'm sharing the install process to help everyone who hasn't yet become friends with linux terminal entirely 
.
Guide to set up Deepstack on the Nvidia Jetson Nano 2GB
1. Getting Started with Jetson Nano 2GB Developer Kit
2. Getting Started with Jetson Nano 2GB Developer Kit
-> install in headless mode
3. log in via ssh and type the following commands
4. if you connect a fan, continue with these commands to auto-controls the fan speed depending on temperature:
first analysis will take 30-60s, then it gets significantly faster
to check deepstack logs:
list all existing containers:
to check ram usage:
.Guide to set up Deepstack on the Nvidia Jetson Nano 2GB
1. Getting Started with Jetson Nano 2GB Developer Kit
2. Getting Started with Jetson Nano 2GB Developer Kit
-> install in headless mode
3. log in via ssh and type the following commands
Bash:
sudo apt update
sudo apt upgrade
sudo docker pull deepquestai/deepstack:jetpack
sudo docker run --runtime nvidia --restart unless-stopped -e MODE=High -e VISION-DETECTION=True -p 80:5000 --name deepstack deepquestai/deepstack:jetpack
sudo systemctl enable docker.service4. if you connect a fan, continue with these commands to auto-controls the fan speed depending on temperature:
Bash:
wget https://github.com/Pyrestone/jetson-fan-ctl/archive/refs/heads/master.zip
unzip master.zip
cd jetson-fan-ctl-master
chmod +x install.sh
sudo bash ./install.shfirst analysis will take 30-60s, then it gets significantly faster
to check deepstack logs:
Bash:
sudo docker logs --tail=50 deepstacklist all existing containers:
Bash:
sudo docker ps -ato check ram usage:
Bash:
free -m
Last edited:
I have several cameras running great. Recently I have added eufy 2k camera. I can install the camera to BI using rtsp but it does not look like I can record or capture image or utilie the camera with the AItool. Is this a shortcoming of rtsp or I am not setting up the camera correctly. I have tried to set up similar to the annke 4k cameras with two streams.
I understand there is only one stream here so I added the eufy twice one to capture the jpeg the other to record. I would appreciate any help. Thanks
I understand there is only one stream here so I added the eufy twice one to capture the jpeg the other to record. I would appreciate any help. Thanks
Hi have problem with ai tools and deepstack when ai tools tries to process an image it give me this error message "ERROR: Processing the image C:\AI-input/altansd.20210321_001024504.jpg failed. Failure in AI Tool processing the image." What can be the problem its not specified in the log why it fails so I need help someone know what to do?
I try to update at last version and getting this error. AITool-VorlonCD-v2.0.988 no such table: History [SQLiteException] (In ) (UpdateHistoryList) is it known for anyone?
I fixed it there was something wrong with deepstack but now it works thanks anyway for trying to help me
I am noticing that my AI triggered video clips are jumping around a bit (i.e. are not a continuous clip). I was under the impression the checking the "Record No Signal and Inactive Frames" was the option to stop the skipping around? Would i also need this "retriggers" option?
I set them all to 3 secs, but now that you mention that I do remember that break time has to play nice with another time setting, the cool down?
There is also a Pre-trigger buffer of 5secs and i take a jpeg on a cloned camera every 4 secs. Though couldn't remember where cool down was right away (or what it did)....maybe it's in AI Tool?