Building an Object Detection Model with the ML.NET Model Builder

Showing how to use the VoTT (Visual Object Tagging Tool) to generate data and use it in Model Builder to get an object detection model.
Stop signs tutorial - docs.microsoft...
VoTT documentation - docs.microsoft...
Feel free to join the ML.NET Community Discord - / discord
ML.NET Playlist - • Create Your First Mach...
Model Builder Playlist - • A Look at the ML.NET M...
If you wish to support my work, feel free to sponsor me on GitHub - github.com/spo...
Contact:
Email: [e-mail]
Twitter: / jwood
Blog: jonwood.co/
Gear used (affiliate links):
Mic - amzn.to/2YEXtxI
Mouse - amzn.to/2ZtASoQ

[@JonWood]

What kind of object detection models do you plan on using Model Builder for?

[@waltergrimm7161]

I would like to use it to detect the progress in welding

[@fromurgench]

Hello dear friend. How I can realise with my webcamera realtime detection? With capture and getting frames no problem, but The InputData only gets Link to my image (ex. c:/asd/asd.png). How I can send Bitmap object to predictor?

[@kon4251]

any idea whether VS plans to enable object detection training for local machines in the near future? I want to use it for small personal/passion projects, and don't really want to spend a dime on Azure

[@ucNguyen-sj7zp]

Thanks for your video. I wonder how to use local environment to train this ?

[@AJ93-lx6on]

what if I had labelled dataset images with Json coco format. will it train directly without labelling.

[@federicocreti4699]

I can't add any compute, I get always this error message: this workspace does not have any valid compute sizes

[@bekmyrzaomorov4123]

Hey bro! Do you solve this problem?

[@spekbalazs6304]

Same here :(

[@federicocreti4699]

I solved downloading last version of model builder plugin extension

[@spekbalazs6304]

@@federicocreti4699 thank you :)

[@bekmyrzaomorov4123]

@@federicocreti4699 Bro! Thanks for answer! Where did u download? In Visual Studio Installer ? Or via link?

[@dangnguyenvan5500]

Great video! i wondering whether is this model can run with real-time? how the FPS? thanks a lot!!

[@Zeeshanmys]

Love your tutorials...!!!! Keep them coming...!!!

[@حيدرعباس-ط9ه7ر]

Hi Jon, I used this tutorial step by step to detect car number plate, but it is very slow, it takes 1.5 to 5 seconds to detect the number plate while the picture is 640×480 pixels! Are there any options that I can use to get it faster? Thank you

[@dangnguyenvan5500]

Dear sir, My visua studio 2019' object detection being in limited scenarios, what i can do to fix this? Thanks in advance!

[@Zeeshanmys]

Hey jon.. Can we make use of this concept to identify a certain text on images..??

[@JonWood]

Depending on what you need Cognitive Services does really well for OCR for general purposes. For a custom model it may be better for image classification similar to MNIST.

[@Anno_Nymous]

if i set up the compute it just errors out.

[@emmanueladebiyi2109]

Nice video

[@Evvox]

Waiting until Microsoft makes it possible to do it local instead of Azure...

[@waltergrimm7161]

your video is broken, I can just see blurred lines