Back to Gap Map Infrastructure / Classic ML & computer vision

Google Cloud Vision API

closed / Overall score: n/a

Google Cloud's hosted image-analysis API, answering REST and RPC calls with Google's pretrained models for labels, object localization, OCR and dense document text, faces, landmarks, logos, image properties and SafeSearch ratings. It is billed per feature per image. Custom models are a separate Google Cloud product, AutoML Vision, rather than part of this API.

Openness

1 high confidence

1.0

license: Proprietary
source: closed(hosted API only

A hosted Google Cloud service: the product page offers the API and nothing to download or run yourself.

https://cloud.google.com/vision recorded 2026-09-26
Product page: "Powered by Google's pretrained computer vision ML models, Cloud Vision API is a readily available API (REST and RPC)"; the page describes only the hosted API.

Adoption

not assessed

Google publishes no usage figure for the Vision API, and it has no package or download count of its own, so no level is assigned.

https://cloud.google.com/vision recorded 2026-09-26
The product page, read for a usage disclosure; it gives pricing and a free tier but no customer or request count.

Capability

4 medium confidence

4.0

The Vision API returns labels, objects, text, faces and moderation ratings for any image with no training step, which is the out-of-the-box end of computer vision. Training a detector for your own objects means moving to AutoML Vision, a separate product, whereas Amazon Rekognition builds custom-label models inside the same service.

https://cloud.google.com/vision/docs/features-list recorded 2026-09-26
Features list: text detection, document text detection, landmark, logo and label detection, image properties, object localization, crop hint detection, web entities and pages, face detection and SafeSearch; object localization points to AutoML Vision Object Detection for customized detection.

Verified 2026-09-26