Article Preview
Buy Now
FEATURE
AI Image Tagger
Using ChatGPT to Label Images
Issue: 23.3 (May/June 2025)
Author: Marc Zeedar
Author Bio: Marc taught himself programming in high school when he bought his first computer but had no money for software. He's had fun learning ever since.
Article Description: No description available.
Article Length (in bytes): 34,925
Starting Page Number: 12
Article Number: 23302
Resource File(s):
project23302.zip Updated: 2025-04-30 22:33:56
Related Link(s): None
Excerpt of article text...
These days, most of my photos are stored in iCloud and are available on my phone. Because Apple Photos can use AI to search through the 20,000 images I've got, I don't really need to organize them that much. I can just search by a person's name, subject, location, or date. It's not perfect, but it works pretty well and is better than my memory.
But that's only true for relatively recent pictures—the last decade or so. I've been using digital cameras since the mid-90s, and I've got lots of old photographs I'd love to sort and categorize.
The problem is that most of those are "organized" into "camera rolls"—essentially just folders of each batch of photos I imported from the camera way back when. Probably half the photos are blurry or too low-resolution to be useful, or they are repeats of the same subject. Sorting through them is a pain.
Isn't this exactly the kind of thing AI should be good for? To me, this is what Apple Intelligence, built into my devices, should be able to do. I want to be able to select a few thousand images and say, "Sort these into categories for me," and end up with folders of sunsets, events, people, places, and so on.
Sadly, it does not work that way... yet. I suspect it will someday. In the meantime, I wanted to explore creating my own tool. Recently, when I was browsing ChatGPT's API (Application Programming Interface) documentation, I discovered that it supports image identification (
https://platform.openai.com/docs/guides/images
). There are some limitations, but it sounded useful. You can upload a photo, and it can describe what's in the image, and I thought perhaps that could be used for categorization.The problem is that the example code is for Python or JavaScript, not Xojo. Those languages already have libraries created by OpenAI to facilitate using their API. Xojo does not.
...End of Excerpt. Please purchase the magazine to read the full article.