A product index with names, colours, sizes and prices, plus a transparent cutout for each product.
About
Takes a folder of catalog page images and turns them into data. It sorts each page (product grid, cover, lifestyle or info), splits product pages into one cell per item by their layout, reads each item's printed name, colour, sizes and price, and cuts each product out onto a transparent background. The result is an index of products plus folders of cutouts and colour swatches.
It is for brands that have a line sheet or catalog as images and need per-product data or product photography out of it. Pages are split by their layout rather than by colour, so pale products on pale backgrounds are still separated. You can run it without cutouts first to check the page split and the text it reads, and existing cutouts are reused unless you ask for new ones.
Every page is read by a vision model and every cutout is a paid call. In Syvon a typical cutout costs about $0.03, so a catalog with dozens of products costs a few dollars.
- TypeScript
- AI vision models
- AI image models
Requirements
- Runs with
- Node.js 20 or later
- Operating systems
- windows, macos, linux
- Network
- Sends page crops to the vision provider and product crops to the background-removal provider over the internet.
Costs
- External costs
- OpenAI: billed to your own key — For reading the pages. Planned direct lane: you would bring your own OpenAI key. Today the call runs through the Syvon provider proxy and Syvon credits.