Imagga Image Recognition API

The v2 image tagging API returns a flat list of descriptive tags, each with a score from 0 to 100. You set a threshold and keep only the tags you trust, which keeps noisy labels out of your search index. Tag labels can be requested in several languages, so a storefront can display keywords in the shopper's language without a second translation step.

The v3 endpoint returns a structured, hierarchical result built for automation rather than a flat keyword list. It suits decision logic and agent workflows where a downstream system needs to know what kind of object it's looking at, not just that a photo contains a "dog". The v3 model exists alongside v2, so older integrations keep working.

One call returns a judgement on whether an image contains adult content, and the same service can flag sensitive text found inside the picture. It's the endpoint most business customers buy first, because it replaces a manual review queue at the point of upload. Thresholds are yours to set, so you can decide how strict the filter is.

Upload a reference image, and the API returns visually similar images from a collection you've indexed. Retail teams use it for "find similar" buttons and for catching duplicate listings. Results come with distance metrics, so you can cut off weak matches instead of showing everything.

The OCR endpoint extracts printed text and returns its position in the image. That covers receipts, packaging, signage, and screenshots. It's text extraction, not document understanding, so expect raw strings plus coordinates and do the parsing on your side.

Smart cropping finds the subject and returns crop coordinates instead of a fixed center cut. Background removal returns a cutout, useful for catalog shots that need a clean white backdrop. A color endpoint returns dominant colors, which helps when you filter or group images by palette.