Wonnx real-time webcam image classication using WebGPU | Hacker News Reader

Wonnx real-time webcam image classication using WebGPU

dl.pixelspark.nl

43 pointsEduard32 comments

It classified me as "chimpanzee, bowtie" which gave me a good laugh, thanks.

+1 for humour. I am a plumber, and my mobile phone is a mini-skirt. Nice combo.

It classified my mostly-bald, short hair as "shower cap". I help a coffee cup up to the camera that it called a plunger. There were some other bathroom-centric classifications as well, maybe it tries to guess the room type and common objects from the scene? FTR, I'm sitting in my open kitchen area, nothing around looks remotely like a bathroom.

I'm a barber shop. Nice to meet you chimpanzee w/ a bowtie!

I can appreciate folks coming to the defense of the demo, but the lede on what makes this special is pretty well buried to folks that aren't in this space.

If there was a blurb on the page stating that the cool bit we should be excited about isn't whether the image classification is accurate, but that there's realtime image classification running against WebGPU. That's definitely neat. However, with no context for folks clicking a random link, I don't think it's really all that off base for folks to comment on the model's functionality being comically inaccurate. At the time of writing, I had to read the bottom comment to get the understanding necessary for why this is neat.

It's fast, but laughably inaccurate. Open shirt? Windsor knot. Hand against chin? Stethoscope. Cell phone close to face? Harmonica.

You should add a button to switch cameras, on my computer it’s stuck on the Open Broadcaster Studio’s virtual camera (even while close) and I’m not sure how to get it use my webcam instead.

It classified me as a punching bag or punching ball, which was pretty funny

Negative, I am a meat popsicle.

Made me think of probably one of the great quotes in sci-fi cinema.

I had some friends in middle school that classified me the same way

People complaining about quality here are missing the point, this is ONNX compatible inference engine written Rust, it just using 5MB SqueezeNet from 2016 for simplicity.

Question is, is it worth to invest time and effort into ONNX ?

https://en.wikipedia.org/wiki/SqueezeNet

https://github.com/onnx/models?tab=readme-ov-file#image-clas...

here is the same model using tensorflowjs

https://hpssjellis.github.io/beginner-tensorflowjs-examples-...

https://t-shaped.nl/posts/running-ai-models-in-the-browser-u...

Missing the point? When the classifications are horribly bad, what is the point? I can write a random phrase generator in FAR less than 5MB that would have the same overall accuracy as this.

The point you're missing that this is a just a demo for Wonnx, an inference runtime for ONNX models.

You can plug your own model into it, it's a general-purpose inference runtime that runs in the browser.

The point is you can swap in any ONNX model, instead of the 5mb shitty one

>"this is ONNX compatible inference engine written Rust"

Ah, the fact that it is written in Holy Rust instantly absolves abysmal quality.

No, the inference accuracy of the image classifier is dependent on the model used and this is a demo of the code executing the model in a browser with GPU acceleration not the model itself. You can plug and play any model in the onnx format e.g. https://github.com/onnx/models. As a comparison, complaining about the "abysmal quality" of the dummy model on display here is like saying blender is bad 3d modeling software after opening it for the first time because all it models is a blank cube.

No the point is it's ONNX and can be pointed to any ONNX model, wow people have really gotten dumb on HN lately

Seriously HN is feeling more like youtube comments section lately I don't know what happened.

https://github.com/webonnx/wonnx-wasm-example

Is this demo running an untrained model?

This has to be the most random and inaccurate image classification I've ever seen.

I don't know... Today I feel like quite the ice lolly. Maybe there's something to it.

It's pretty sure that I'm a barbell and my xbox controller is a sewing machine.

I think this is just spitting out random text. April the first is not here yet...

Car mirror, barbershop, cowboy hat

I get it, it's an XKCD password generator https://xkcd.com/936/

Got: >panicked at 'No GPU found given preference', /Users/tommy/Git/wonnx/wonnx/src/resource.rs:13:10

Is the path to the rust hard coded?

Rust includes source file and line information in many error messages for easier debugging—it's not looking for anything at that path.

So I'm a ping-pong ball, good to know

Apparently I'm a bathing cap, my keyboard is blowdryer, my mouse a dumbbell.

That's pretty impressive

Just seemed to classify everything as either mob, swab, hairspray or barbell

I am a 'washing machine' and my hand is a 'spatula'

apparently my office is a barbershop. That explains so much

same :D

Reply on news.ycombinator.com