Hi! I’ve read all of the previous posts regarding this same topic, however Sofia’s solution (which is amazing!!) is a bit too complex for a beginner like me. I was wondering if there’s an easier way to implement my AI to thunkable. Are there alternatives I could do? Should I just use the web viewer? Thanks for the help!
Welcome to Thunkable!
Can you post a link to “Sofia’s solution”? It’s not clear what you’re wanting. Are you trying to train an AI? You can do that with a third-party product but it’s not built-in to Thunkable. You’ll probably need to access whatever you use with the Web API blocks in Thunkable.
If you give an example, someone here might have more ideas for you.
Thank you!
Sure! Here’s her article: Sofia Galan’s solution
I already have my Teachable Machine ready. I want to make my app use the teachable machine in order to let the user know if the trash in the image that they took is recyclable or not. Sofia’s app is almost exactly what I am going for:
I’ll notify @Sofia here but I’m not sure she’s still active on the forums. She’d be the best person to ask about this.
I can tell you from experience training an image recognition AI that it’s complicated work. You can read about my attempts here: Image Recognition & Classification using Artificial Intelligence. It took me hours of trial and error and reading documentation to get it right. And even after all that, I couldn’t afford to pay for what I really needed to make an app.
The built-in image recognition in Thunkable is pretty basic but you could start with that and see if it’s enough for you: Camera Blocks - Thunkable Docs.
Hi!
When I made that tutorial, teachable machine was new and the only solution I found was to create that api myself. Deploying an api solution like that it’s hard work haha not going to lie this is the reason I’m in software now.
But I’ve found maybe something easier for you. Sadly you need something that could read that tensorflow JavaScript file. Thunkable is not able to do that at the moment. But there is another tool I use and I use it mainly when I want to launch something fast without dealing with deployments → Glitch.
Glitch it is a code base tool unlike Thunkable or Teachable Machine. You don’t need to worry about infrastructure but you do need to do some little code to make it work. I would try something like this → Tensorflow Js Image Classification and mix it up with a basic API template from glitch to create an API so you can use the rest api blocks in Thunkable. 
So at the end, the solution still needs some code.
Hope this helps!
Sofia
2 Likes
Thanks so much Sofia! I appreciate your reply and I’ll definitely give it a try. Thanks for the help!
1 Like