bi-aidetection VS argos

I run Blue Iris with https://github.com/VorlonCD/bi-aidetection (getting very long since the last update, but still works great) and am very happy with it and am using DeepQuest AI's latest Windows release rather than Docker. I have 14 cameras going to a VM running it all with 24/7 low res recording and event triggered high res recording. I have MQTT reporting of events to Home Assistant for notifications and Blue Iris handles all the actual video pieces. The hardest part is finding decent RTSP cameras that don't re-stream from the cloud. I've got a mix of Amcrest, Reolink, and Unifi. The older Amcrest seem to work best, but the Unifi ones (once you get past the restream from the local UMD bit) are really good as well and provide a nice double layer for notification if I ever want it. I've been running this setup for just a little over 3 years now and the only real complaint I have is about the Reolink cameras. I have them working, but do not recommend them at all.

and this tool to gain ai detection https://ipcamtalk.com/threads/tool-tutorial-free-ai-person-detection-for-blue-iris.37330/

​ https://github.com/VorlonCD/bi-aidetection

1) https://github.com/gentlepumpkin/bi-aidetection 2) The fork: https://github.com/VorlonCD/bi-aidetection

This is a thread that discusses a technique for analyzing still images coming out of Blue Iris using DeepStack initially and later DeepQuestAI. Between those three links you should be able to get going.

The easiest way to think of it is a camera aimed at a driveway. You are probably interested in people and cars in a driveway. Since DeepStack doesn't know which part of the frame has motion, it will alert on a parked car in the frame even if something else triggered the motion. Dynamic masking will automatically apply an "ignore" mask on an object that has remained stationary for several frames and it will keep it ignored until it moves again. This allows you to enable alerts for cars and people, and only get alerted when a new car arrives in the driveway, or an existing parked car moves. This is the fork of AI_Tool that I use to accomplish this: https://github.com/VorlonCD/bi-aidetection/releases

I think what you are asking for came from this project: GitHub - VorlonCD/bi-aidetection: Alarm system for Blue Iris based on Artificial Intelligence.

@diymatt https://github.com/VorlonCD/bi-aidetection cool deepstack implementation for blue Iris

you say you have followed argos installation instructions uptil the point of creation of the systemd services. bear in mind that you only need the argos_serve for argos-presence not the argos_stream service. disable and remove the argos_stream service if you started that as well.

I then entered the commands from the "Installation" section from this page: https://github.com/angadsingh/argos#installation . I also entered the 3 commands from the tf2 section. Then the commands to make a systemd service to run it automatically.

it will only require a few line of changes for me to add support for coral (in the way the tensorflow lite API is initialized), however I do not have a coral accelerator with me to test. Appreciate if you can please open a github issue for this and docker so that I can work on it when I get time? https://github.com/angadsingh/argos/issues

the test is running the videos at 10 fps for the purpose of the demo video, otherwise they run at 100+ fps easily. config for this test is here