How to Start Using Machine Learning in a Simple iOS App?

General Discussion
1

Hey everyone,

I’m still learning iOS app development (thanks to Chris’s awesome videos!),
image
and I’m curious about adding something basic from machine learning into a simple app—nothing too complex, maybe just image recognition or some kind of smart suggestion.

I’ve seen a few resources online, but many feel too advanced or assume I already know how ML works. Is there a beginner-friendly Machine Learning Tutorial you’d recommend that also connects well with Swift or iOS frameworks like CoreML?

I’m building a hobby project—a photo journal app—and I thought it’d be cool if it could recognize simple things in images like “cat” or “food” to auto-tag entries. Just wondering if that’s too ambitious for a beginner or doable with the right guide.

Would love any advice, links, or even small code examples if someone has tried something similar.

Thanks in advance for any help!
-NicoleS.

Awaiting for a SHOUTOUT from chris :slight_smile:

2

Hey Nicole!

That sounds super cool, I’ve been looking into something like this for a different app idea, and one of the things I came across was Apple’s Vision framework.

I did a quick dive into the documentation and it looks like it’s a really useful framework and pretty straight forward to access. As far as labeling all the things in a photo, it might be easier to niche down to a few things to start. The quality of the photo they take will affect the reliability a lot too. Still, if you’re interested in doing it, it’d be fun to learn and totally worth trying.

For example, depending on what kind of image recognition you’re thinking about, they have a built-in request for detecting animals in photos, which could be really useful for that cat label you mentioned:

I’m definitely not an expert, but I hear a lot about this stuff from my husband — he’s a perception/robotics engineer, so it’s always kind of floating around in the background at our house as he’s very passionate about what he does :smile:.

I have a friend who integrated with an API to label certain photos for her app after a 3-month SWE bootcamp. It was python and she mentioned most of the ML stuff was abstracted away, she really enjoyed it.

Anyhow, good luck it seems like a fun idea! I am super curious to know how it goes.