Teachable Machine

5/20/2025 at 4:32:38 PM

Could this be used to determine whether or not a food item is a hot dog?

by woah

5/20/2025 at 4:39:21 PM

Or differentiate between a chihuahua and a muffin?

https://www.freecodecamp.org/news/chihuahua-or-muffin-my-sea...

by eitally

5/20/2025 at 5:09:40 PM

Yes

by lxe

5/20/2025 at 8:52:43 PM

Like Shazam for food

by blitzar

5/20/2025 at 5:16:10 PM

I believe this is from 2019 or 2020. Any LLM e.g. Gemini could one shot any problem here, I believe.

by varunneal

5/20/2025 at 8:20:11 PM

It's useful for teaching kids about machine learning. Actually, IMO it's the only tool for doing so that isn't complete garbage. Everything else in the space is terrible.

by Wowfunhappy

5/21/2025 at 1:25:54 PM

I have used it for just this purpose with kids as early as fourth grade with a lot of success, actually!

by ChicagoBoy11

5/22/2025 at 2:25:34 PM

To be clear, Teachable Machine is great. It's everything else that is awful.

by Wowfunhappy

5/20/2025 at 6:36:47 PM

correct, youtube video shows it was uploaded 5 years ago - doubt it is relevant today even if someone don't want to use VLM for those tasks. Google dropped ball on tensorflow lite and their rebrand to LiteRT is still WIP.

Probably safer bet is to use onnxruntime or executorch if someone needs to deploy on edge devices today. At least for onnx, community is huge on hugging face and plenty of modern SOTA models already in transformers and transformers.js.

by pzo

5/20/2025 at 5:33:02 PM

[dead]

by wetpaws

5/20/2025 at 5:34:42 PM

I use it to explain machine learning to kids. You can integrate it in a custom version of scratch: https://playground.raise.mit.edu/create/

by puzu

5/20/2025 at 9:34:43 PM

I agree that it is a useful tool for teaching students about ML.

I'm going to use it in a course to teach a machine to recognize claps vs finger-snaps.

If you know of other simple, light tools like this that can be used to teach / demonstrate ML, please share.

Thank you.

by profsummergig

5/20/2025 at 4:34:33 PM

Curious if you did make something can you export it into some format and use it in like an RPi somehow (provided the inference power isn't crazy). I have used a Pico Voice model before for example set it up on a computer, brought it into a Pi and used it there with a mic.

Edit: I do see languages in there like NodeJS so yeah must be

"export your model" TF.js ahh yeah nice

by ge96