Google Cloud Vision API Product Search

What Google Cloud Vision API Product Search is

Google Cloud Vision API Product Search helps retailers search product catalogs with images. Retailers create products with reference images, arrange them in product sets, and submit a query image to find visually and semantically similar products. Supported categories are homegoods, apparel, toys, packaged goods, and general. Query images can be provided from Google Cloud Storage, a web URL, or as a base64-encoded string. Search requests can use optional product-label key-value filters with AND or OR expressions, subject to expression rules.

The API is available through REST or RPC, with Google Cloud client libraries for supported programming languages. A CSV bulk import can create product sets, products, and reference images together. The product index updates approximately daily, so additions and deletions may take until the next update to appear in results. Each image file can be up to 20 MB; JSON request objects can be up to 10 MB, and a request can contain up to 16 images. A product can belong to no more than five product sets. The free plan covers prediction and storage at 0–1,000 images per month. Listed paid rates include $0.10 per 1,000 images for storage and prediction rates that vary by monthly image volume.

Who it suits

It suits retailers who want to search their product catalogs using images and need visual and semantic similarity search. It is also a fit for teams that can work with an API, client libraries, or CSV bulk import.

Limits

The free plan covers prediction and storage for 0–1,000 images per month. Listed storage costs $0.10 per 1,000 images for 1,001–20,000,000 images/month; prediction is $4.50 per 1,000 images for 1,001–5,000,000 images/month and $1.80 per 1,000 images for 5,000,001–20,000,000 images/month. Requests allow up to 16 images, with a 20 MB per-image and 10 MB per-JSON-request limit; indexing updates approximately daily.

Getting started

1. Open the Google Cloud Vision Product Search documentation.
2. Create products with reference images and organize them into product sets, or prepare a CSV bulk import.
3. Choose REST, RPC, or a Google Cloud client library for supported languages.
4. Submit query images from Cloud Storage, a web URL, or a base64-encoded string.
5. Apply optional product-label filters if needed.

Verdict

Choose Google Cloud Vision API Product Search if you need an API-based way to match query images against a retailer product catalog. Its filters, bulk import, and supported input sources offer several ways to prepare searches and catalog data. Look elsewhere if your workflow depends on immediate index updates or more than five product sets per product.