google_vision_cli 2.0.1+1

A CLI tool for making API requests to Google Vision API using the google_vision package.

example/README.md

Landmark Detection Example

This example demonstrates landmark detection using the vision CLI on an image of the CN Tower in Toronto, Canada.

Prerequisites

• Google Cloud service account with the Vision API enabled
• A JSON credentials file for the service account

Setup

Place your service account credentials file at the default path:

mkdir -p ~/.vision
cp /path/to/your/credentials.json ~/.vision/credentials.json

Or pass a custom path with --credential-file.

Run Landmark Detection

Detect landmarks in the image:

dart run vision detect --image-file example/cn_tower.jpg --features LANDMARK_DETECTION

Expected Output

A successful response returns an entity annotation with the landmark name, confidence score, and bounding polygon:

{
  "mid": "/m/0c6m",
  "description": "CN Tower",
  "score": 0.987,
  "boundingPoly": {
    "vertices": [
      { "x": 120, "y": 80 },
      { "x": 500, "y": 80 },
      { "x": 500, "y": 720 },
      { "x": 120, "y": 720 }
    ]
  }
}

Additional Detection Types

The same image supports other Vision API features. Try them together:

Label Detection

dart run vision detect --image-file example/cn_tower.jpg \
  --features LANDMARK_DETECTION,LABEL_DETECTION

Returns both the landmark name and general labels (e.g., "skyline", "city", "tower", "landmark").

Safe Search

dart run vision detect --image-file example/cn_tower.jpg \
  --features SAFE_SEARCH_DETECTION

Returns likelihood ratings for adult, spoof, medical, violence, and racy content.

Crop Hints

dart run vision crop_hints --image-file example/cn_tower.jpg

Returns suggested crop region vertices for the image.

Run from the Project Root

From the repository root:

dart run vision detect --image-file packages/google_vision_cli/example/cn_tower.jpg \
  --features LANDMARK_DETECTION