Vision AI / Video Intelligence Complete Guide: Image & Video Analysis APIs

GCP's image and video AI is built around three services — Cloud Vision API (image analysis),Video Intelligence API (video analysis), and Vertex AI Vision (custom models + edge). This article summarizes the features of each, when to use which, and how they compare to AWS and Azure.

Cloud Vision API Features

Vision API Usage Example

from google.cloud import vision

client = vision.ImageAnnotatorClient()
image = vision.Image()
image.source.image_uri = "gs://my-bucket/photo.jpg"

# 複数機能を 1 リクエストで
features = [
    vision.Feature(type_=vision.Feature.Type.LABEL_DETECTION, max_results=10),
    vision.Feature(type_=vision.Feature.Type.TEXT_DETECTION),
    vision.Feature(type_=vision.Feature.Type.SAFE_SEARCH_DETECTION),
]
request = vision.AnnotateImageRequest(image=image, features=features)
response = client.annotate_image(request=request)

for label in response.label_annotations:
    print(f"{label.description}: {label.score:.2f}")
print("OCR:", response.text_annotations[0].description)
print("Adult:", response.safe_search_annotation.adult.name)

Video Intelligence API Features

• Label Detection: Video-, scene-, and frame-level labels
• Shot Change Detection: Scene change points
• Object Tracking: Object tracking inside the video
• Text Detection: OCR for text inside the video
• Logo Detection: Brand logo tracking
• Explicit Content: Unsafe content detection
• Speech Transcription: Transcription of speech in the video
• Person Detection: Person detection plus pose estimation
• Face Detection: Face detection (not identification)

Vertex AI Vision

• Custom model training (image classification / object detection / OCR)
• Stream Ingestion (ingest video streams from RTSP / Pub/Sub)
• People Counting / Occupancy Analytics (for retail and facilities)
• Person/Vehicle Detection (for surveillance cameras)
• IoT edge deployment (Coral / Jetson)
• BigQuery integration for large-scale analytics

Pricing Examples

Cross-Cloud Comparison

Typical Use Cases

• E-commerce: auto-tagging product images and blocking inappropriate images
• Manufacturing: visual inspection (Vertex AI Vision custom models)
• Retail: in-store foot-traffic analytics (Vertex AI Vision Occupancy)
• Media: video content moderation and metadata generation
• Logistics: shipping label OCR
• Security: anomaly detection on surveillance cameras

Note: Google Cloud is a trademark of Google LLC. For the latest information, see the official Cloud Vision documentation.