IP Cam Talk Custom Community DeepStack Model

[icpilot]

You can try disabling DeepStack default objects model using "objects:0" when using the custom models

Yep, I planned to do that. It is done in each camera config, right? I never made it that far till the Rekor Agent crashed.

 

[MikeLud1]

It is done in each camera config, right?

Yes

 

[icpilot]

Yes

No matter. I will sort it out later. Your project is really taking shape and I was hoping to contribute, but I will when I can.

Carry on .... I'll catch up with you once I sort out my issue.

 

[MikeLud1]

Microsoft Remote Desktop started crashing with an Error code: 0x112f

Search google there are several results on the fixes

 

[MikeLud1]

General V1.1 just added in the first post

I figured out how to download the images that DeepStack used to train the object model. With these images I can start to make the community custom DeepStack model, below are the steps to create the custom model.

The first step would be to create a new DeepStack custom model using the same image that DeepStack used but removing all the labels we do not want. Below is a list of the labels that I think we should keep and a list of labels we should remove. Please let me know if I should change the lists, I am planning to create the new model this weekend so let me know by Friday if I should change the list.

The second step would be to start adding new labels with images the everyone contributes to the custom model.

Combined Labels	General Labels	Animal Labels	Labels removed from Deepstack original model
person, bicycle, car, motorcycle, bus, truck, bird, cat, dog, horse, sheep, cow, bear Future Labels: deer, rabbit, raccoon, fox, coyote, possum, skunk, squirrel, boar	person, bicycle, car, motorcycle, bus, truck	bird, cat, dog, horse, sheep, cow, bear Future Labels: deer, rabbit, raccoon, fox, coyote, possum, skunk, squirrel, boar	airplane, train, boat, traffic light, fire hydrant, stop sign, parking meter, bench, elephant, zebra, giraffe, backpack, umbrella, handbag, tie, suitcase, frisbee, skis, snowboard, sports ball, kite, baseball bat, baseball glove, skateboard, surfboard, tennis racket, bottle, wine glass, cup, fork, knife, spoon, bowl, banana, apple, sandwich, orange, broccoli, carrot, hot dog, pizza, donut, cake, chair, couch, potted plant, bed, dining table, toilet, tv, laptop, mouse, remote, keyboard, cell phone, microwave, oven, toaster, sink, refrigerator, book, clock, vase, scissors, teddy bear, hair dryer, toothbrush

​

Combined V1.1 Training Results:​

View attachment 111736View attachment 111739​

General V1.1 Training Results:​

View attachment 111835View attachment 111836​

 

[105437]

Currently reviewing and labeling 4,000 images that I have captured for the month of December. Will take me a few days to complete and send.

 

[sebastiantombs]

Is the objects list still the same in general V1.1?

 

[MikeLud1]

Is the objects list still the same in general V1.1?

Yes, when the objects change I will make it V2.0

 

[105437]

@MikeLud1 It actually went faster than I thought. There were several images in my capture folder that didn't need labels. The image folder and label file are compressing right now. What is the size limit for an attachment to a DM?

 

[MikeLud1]

What is the size limit for an attachment to a DM?

I am not sure, try DMing the zip file and see if it lets you.

 

[MikeLud1]

@MikeLud1 It actually went faster than I thought.

Once you start using Make Sense it's easy to label the images

 

[sebastiantombs]

@MikeLud1 Do you want the images segregated into day and night categories or all in one file?

 

[MikeLud1]

@MikeLud1 Do you want the images segregated into day and night categories or all in one file?

It will help, because I need to have 5% of day/night for validation and 95% of day/night for training.

 

[sebastiantombs]

OK, I'll trash what I started and do it over again for day and night. That dawned on me as I was fooling with tagging the images. Just trying to make it easier for you to handle is the goal.

 

[105437]

@sebastiantombs How are you creating the source label images?

 

[sebastiantombs]

I'm using the list from the combined model posted by Mike.

 

[105437]

I'm using the list from the combined model posted by Mike.

Should have been clearer. How are you getting the jpg files?

 

[MikeLud1]

Should have been clearer. How are you getting the jpg files?

What you can do is make a clone camera and only have it using motion detection (No DeepStack or 24x7 recording) just have the camera alert on motion.

 

[Tinman]

Just for your information I just tested the combined 1.1 CPU version Here is before I added it:



And after installing and using the combined.pt

 

[105437]

What you can do is make a clone camera and only have it using motion detection (No DeepStack or 24x7 recording) just have the camera alert on motion.

I used saved IVS images directly from the cameras.