People can mask USPS AI detections

[AlphaBlueIris]

Not sure why it is not showing under alerts, what am i doing wrong? Thank you.

 

[VideoDad]

In my case, I have a master camera set up for recording any vehicles, people, etc. from that camera. I have a cloned camera (called "Delivery") where I set-up simple motion detection (no zones, etc.) and have it looking for the desired delivery logos:


The email alert settings are pretty straightforward too:

 

[AlphaBlueIris]

Thank you. I did the same thing you did, not sure why mine does not trigger the alert as in the picture I attached before. Everything looks correct, it did find the FedEx truck. What are the rules, that is were I'm confused about. From that picture the delivery was found all the time, the license not but I did not include it in the custom objects.

Sent from my SM-S906U using Tapatalk

 

[VideoDad]

Thank you. I did the same thing you did, not sure why mine does not trigger the alert as in the picture I attached before. Everything looks correct, it did find the FedEx truck. What are the rules, that is were I'm confused about. From that picture the delivery was found all the time, the license not but I did not include it in the custom objects.

Sent from my SM-S906U using Tapatalk

What are your motion sensor settings? I don't have any object or zones enabled to make it as simple as possible. Also, it seems you processing 8 images x 750ms (3/4 second) = 6 seconds. So are you sure that from the time BI senses motion until the logo is clearly in frame, it only takes 6 seconds?

 

[AlphaBlueIris]

The way I had it configured was with the motion detection with Zone A and G. Only when the person deliver the package it will look for the delivery truck. I increased the time as you said to 15 seconds, hopefully it helps. Thanks again.

 

[VideoDad]

The way I had it configured was with the motion detection with Zone A and G. Only when the person deliver the package it will look for the delivery truck. I increased the time as you said to 15 seconds, hopefully it helps. Thanks again.

Ok, I see what you are trying to do. If you see any motion (shadows, dog, person, whatever) in your lawn/walkway area, then check if a delivery logo is visible and flag/alert on the clip.

I'm not sure it will work that way given the way BI processes the AI objects. I have a feeling it sees that the FedEx logo is there but it might see it as an existing, non-moving object.

It does look like the custom model is detecting it correctly, so that's good. I'm just not sure on the ins and outs of BI to set this up the way you want. I know I had problems previously when I was trying to do different AI detections in separate zones but worked around my issues with additional cloned cameras.

Anyone else have ideas?

 

[AlphaBlueIris]

What I fail to understand is why it sees the FedEx at every frame but fails to mark it. I guess I do not understand well how it get triggered. I know it is on the zone g and it sees it as zone g. By the way, you did a great job with the logos, it got close to 90%.

Sent from my SM-S906U using Tapatalk

 

[JL-F1]

Any chance of getting an ONYX version of this so I can use it with Code project .net?

 

[MikeLud1]

Any chance of getting an ONYX version of this so I can use it with Code project .net?

I just made one, below is the download link

https://github.com/MikeLud/CodeProject.AI-Custom-IPcam-Models/raw/main/ONNX%20models/delivery.onnx

 

[JL-F1]

Thank you sir!

 

[David L]

Ok, I see what you are trying to do. If you see any motion (shadows, dog, person, whatever) in your lawn/walkway area, then check if a delivery logo is visible and flag/alert on the clip.

I'm not sure it will work that way given the way BI processes the AI objects. I have a feeling it sees that the FedEx logo is there but it might see it as an existing, non-moving object.

It does look like the custom model is detecting it correctly, so that's good. I'm just not sure on the ins and outs of BI to set this up the way you want. I know I had problems previously when I was trying to do different AI detections in separate zones but worked around my issues with additional cloned cameras.

Anyone else have ideas?

Being still very new to all of this AI, even though I have been running AI when DeepStack was included in BI and now run CodeProject.AI (v2.08-Beta), I have a few questions on these custom models.

I have ran clones before of my cameras, running one 24/7 and the clone detecting motion/alerts. You mentioned running a clone, is this just for your delivery model or do you include other custom models like @MikeLud1 offers in the same clone?

BTW, you guys rock, thank you for all you are doing for the Community...

 

[VideoDad]

Being still very new to all of this AI, even though I have been running AI when DeepStack was included in BI and now run CodeProject.AI (v2.08-Beta), I have a few questions on these custom models.

I have ran clones before of my cameras, running one 24/7 and the clone detecting motion/alerts. You mentioned running a clone, is this just for your delivery model or do you include other custom models like @MikeLud1 offers in the same clone?

BTW, you guys rock, thank you for all you are doing for the Community...

Personally I like to separate the function of the delivery detection to make it easier to send just those alerts, but I don't see why it wouldn't work combined. For me it is easier to deal with issues, say if the detection of another object is causing conflicts.

 

[Vettester]

I have ran clones before of my cameras, running one 24/7 and the clone detecting motion/alerts. You mentioned running a clone, is this just for your delivery model or do you include other custom models like @MikeLud1 offers in the same clone?

Typically a cloned camera has a specific purpose that uses a single model. For example, my driveway camera is setup to detect people, cats and dogs using the ipcam-combined model 24 hrs a day. I have also cloned the camera to use the USPS model that captures the mailman's vehicle, but only between sunrise & sunset and only if the vehicle is traveling from west to east. This triggers my home automation system to let me know the mail has arrived. A second clone of the same camera is setup to count the number of vehicles that drive by my house using ipcam-general model. I knew there was a lot of traffic on my street, but I had know idea how bad it was until I started counting the cars.

 

[David L]

This is really kewl. I just recently got Home Assistant setup. I was lazy and used SmartThings then Hubitat. I added my first automation to have a snapshot sent to our phones if someone presses our doorbell button or motion is detected. So now my focus is on AI more. @Vettester I like your setup. Man you have a lot of traffic on your street, we get a lot too, it is a through street for the front and back part of our subdivision.
One thing, we have Community Mailboxes so don't know if I can capture USPS unless they deliver a package to our front door. Those are our Mailboxes in the distance...

 

[David L]

So Noob question, how do I add the delivery.pt file to CP.AI? To what folder/directory? Is there an Install Link? Appreciate it...

This is what I am running, runs great on my 970 card...Not doing any License plates yet...

 

[Skinny1]

Is there a package.onnx file for the front porch?

 

[MikeLud1]

So Noob question, how do I add the delivery.pt file to CP.AI? To what folder/directory? Is there an Install Link? Appreciate it...

This is what I am running, runs great on my 970 card...Not doing any License plates yet...
View attachment 161066View attachment 161067

For the Object Detection (YOLOv5 .NET) module you need .ONNX model, you can download the delivery.onnx model from the below link. The models need to saved in the below folder.

CodeProject.AI-Custom-IPcam-Models/ONNX models at main · MikeLud/CodeProject.AI-Custom-IPcam-Models

Contribute to MikeLud/CodeProject.AI-Custom-IPcam-Models development by creating an account on GitHub.

github.com

 

[MikeLud1]

Is there a package.onnx file for the front porch?

Yes, download it from the below link

CodeProject.AI-Custom-IPcam-Models/ONNX models at main · MikeLud/CodeProject.AI-Custom-IPcam-Models

Contribute to MikeLud/CodeProject.AI-Custom-IPcam-Models development by creating an account on GitHub.

github.com

 

[Vettester]

@MikeLud1, how about converting this to an onnx file?

Release v1.1 · sstratoti/DeepStack_USPS

Re-ran the training using 1,000 epochs instead of the default of 300. Should be more accurate.

github.com

 

[David L]

For the Object Detection (YOLOv5 .NET) module you need .ONNX model, you can download the delivery.onnx model from the below link. The models need to saved in the below folder.

CodeProject.AI-Custom-IPcam-Models/ONNX models at main · MikeLud/CodeProject.AI-Custom-IPcam-Models

Contribute to MikeLud/CodeProject.AI-Custom-IPcam-Models development by creating an account on GitHub.

github.com

View attachment 161068

Thank You Mike, sorry for Noob question. I was almost there since I searched for .pt and found them in the Yolo folder. Thank you for correcting my mistake since I now run .NET



Hey Mike, you might want to create a signature with Noob info, that is a link to your GitHub page. I went overboard on my IPCAMTALK Profile signature

What I did is include a link to our Doorbell 101 page here, your signature could point to:

CodeProject.AI Version 2.0

CodeProject.AI version 2.0 was released Jan 16, 2023, this thread is for all topic CodeProject.AI version 2.0. https://www.codeproject.com/Articles/5322557/CodeProject-AI-Server-AI-the-easy-way If you need any info on CodeProject.AI version 1.x please refer to the below link (there is some...

ipcamtalk.com

Where you could keep updates like we did in our 101...