I had the opportunity to develop a plugin for Strapi, in the same vein as the plugin I developed for Obsidian.
So I had to immerse myself in Strapi's plugin architecture.
The plugin's front-end is in React, while the back-end is in classic Node.js with Strapi's internal library.
After understanding how Strapi plugins work, I created this plugin by setting up the various back-end routes for analyzing an image.
The principle is simple: I take a list of my images, send them to the back-end, then analyze each image with GPT Vision. Then, with another call to a GPT 3.5 template this time, I format and process all the information analyzed by GPT Vision.
Finally, I directly modify the alternative text, caption and title of the image.
I've also created a simple user interface so that I can follow the progress of image processing and get feedback on what's going on, as well as a small gallery for viewing images in progress.
