Google Cloud Platform (GCP) Cloud Vision API Overview
Google Cloud Platform (GCP)
Used for working with and recognising content in still images
GCP Cloud Vision documentation
Usage Pattern (versus AutoML Vision)
Objective - enable ML practitioners to harness the power of Google's ML for images
Primary use case - face detection, OCR, object detection, etc
Data requirements - just images (labelled or not)
Output format - as required
Customisation - can be customised
Efforts - high for end-to-end model development
Features
Label Detection - broad sets of categories
Face Detection
Features
Multiple faces
Key facial attributes - "faceAnnotations"
Emotional state (surprise, anger, joy & sorrow likelihoods)
Head wear likelihood
Angles: roll, pan, tilt
Detection confidence
Under exposed likelihood
Blurred likelihood
Bounding polygon of face
Web Annotations - similar images from the Internet: entityId, score, description
Optical Character Recognition (OCR) - text detection, support for multiple languages
Explicit Content Detection - adult or violent content
Landmark Detection - popular natural or man-made structures
Logo Detection - popular product logos