IP Cam Talk Custom Community DeepStack Model

[MikeLud1]

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.

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

Thanks to @105437 for contributing images.

Combined Labels	General Labels (includes dark models images)	Animal Labels	Dark Labels	Labels removed from Deepstack original model
person, bicycle, car, motorcycle, bus, truck, bird, cat, dog, horse, sheep, cow, bear, deer, rabbit, raccoon, fox, skunk, squirrel, pig Future Labels: coyote, possum	person,vehicle	bird, cat, dog, horse, sheep, cow, bear, deer, rabbit, raccoon, fox, skunk, squirrel, pig Future Labels: coyote, possum	Bicycle, Bus, Car, Cat, Dog, Motorcycle, & Person	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 V2.0 Training Results:​

​

​

General V3.1 (includes dark models images) This model was trained by @Futaba Training Results:​

​

​

​

Animal V2.0 Training Results:​

​

​

Dark V1.0 Training Results:​

​

​

 

[wittaj]

Wow - taking all that out should make it a lot faster. It is fun to try out the toothbrush LOL, but if removing 2/3rds of the objects speeds it up, that would be wonderful.

I would almost like to see two custom models - one for vehicles/humans and one for animals.

Labels To Keep:
person
bicycle
car
motorcycle
bus
truck

Animal Labels:
bird
cat
dog
horse
sheep
cow
bear

 

[MikeLud1]

Wow - taking all that out should make it a lot faster. It is fun to try out the toothbrush LOL, but if removing 2/3rds of the objects speeds it up, that would be wonderful.

I would almost like to see two custom models - one for vehicles/humans and one for animals.

Labels To Keep:
person
bicycle
car
motorcycle
bus
truck

Animal Labels:
bird
cat
dog
horse
sheep
cow
bear

That can be done, may have one with all the labels and then two separate models one for vehicles/humans and one for animals.

 

[wittaj]

That would be awesome!

I look forward to your work!

 

[MikeLud1]

That would be awesome!

I look forward to your work!

The second phase would be for everyone to contribute images for new labels for example raccoons.

 

[sebastiantombs]

How about adding deer, rabbit, raccoon, fox, possum? I have, or can get, shots of all particularly at night. Just say when.

 

[samplenhold]

Wait a minute...you are going to remove broccoli?

 

[MikeLud1]

How about adding deer, rabbit, raccoon, fox, possum? I have, or can get, shots of all particularly at night. Just say when.

Let me finish the first phase then I will post instructions on how the prepare the images with the labels.

 

[MikeLud1]

Wait a minute...you are going to remove broccoli?

I guess I can add broccoli back, not

 

[MikeLud1]

Below are samples of the image used for the model

 

[MikeLud1]

Wait a minute...you are going to remove broccoli?

I hope @IReallyLikePizza2 doesn't find out I am removing pizza.

 

[MikeLud1]

Hopefully when I train the custom model I do not blow up my GPU, when I trained the custom LPR model it tool just under 2 hours to train 1.5k of images. This new model will use 84.5k of images.

 

[MikeLud1]

What should I name the new custom models files?

 

[sebastiantombs]

Uhhh, how about general and animules?

 

[105437]

Really appreciate your time to do these models for the community! When you're ready, I can contribute images of deer, fox, raccoon, skunk etc. I get them occasionally. I'm assuming you would want daytime images, right?

 

[MikeLud1]

Really appreciate your time to do these models for the community! When you're ready, I can contribute images of deer, fox, raccoon, skunk etc. I get them occasionally. I'm assuming you would want daytime images, right?

Day and night images would work.

 

[sebastiantombs]

You want night captures of vehicles with headlight bloom? I can supply tons of them!

 

[MikeLud1]

You want night captures of vehicles with headlight bloom? I can supply tons of them!

Is DeepStack currently identifying the vehicles correctly, if so then not needed. If not I can add the images for the training.

 

[sebastiantombs]

It's kind of hit and miss with more miss than hit. I've got HLC up around 80 or 85 and it's better but still misses a lot. This is monitoring street traffic not an LPR.

 

[MikeLud1]

Moved Custom Model to first post

Since I did not see any objections to my selection of labels to keep I completed training the fist model. I did a quick test with the new model and the model looks to be working about the same. I are still validating if the new model has quicker times and better accuracy. Below are the new labels and BI AI settings. Please let me know how the new model works for you.

Labels for Combined model
person
bicycle
car
motorcycle
bus
truck
bird
cat
dog
horse
sheep
cow
bear