Comparison

Google Cloud Vision SafeSearch: API, Pricing & Limits

Google Cloud Vision SafeSearch returns likelihood scores for adult, spoof, medical, violence, and racy content at per-image pricing. Vettly turns those signals into product decisions with policies, actions, dashboards, and appeals - and adds text and video.

What Google Cloud Vision SafeSearch does

SafeSearch Detection is a feature of the Cloud Vision API. You send an image (base64 or a Cloud Storage / public URL) to the images:annotate endpoint with the SAFE_SEARCH_DETECTION feature, and it returns a safeSearchAnnotation with one likelihood per category.

It does not return a decision. Each category comes back as UNKNOWN, VERY_UNLIKELY, UNLIKELY, POSSIBLE, LIKELY, or VERY_LIKELY, and your code decides which combinations to allow, blur, send for review, or block.

• • adult - nudity and sexual content
• • racy - suggestive content short of explicit (swimwear, revealing clothing)
• • violence - violent or gory imagery
• • medical - medical or surgical imagery
• • spoof - memes or edits that alter an original image to make it offensive or funny

curl -X POST \
  -H "Authorization: Bearer $(gcloud auth print-access-token)" \
  -H "Content-Type: application/json" \
  https://vision.googleapis.com/v1/images:annotate \
  -d '{
    "requests": [{
      "image": { "source": { "imageUri": "https://example.com/upload.jpg" } },
      "features": [{ "type": "SAFE_SEARCH_DETECTION" }]
    }]
  }'

# Response
{
  "responses": [{
    "safeSearchAnnotation": {
      "adult": "VERY_UNLIKELY",
      "spoof": "UNLIKELY",
      "medical": "VERY_UNLIKELY",
      "violence": "UNLIKELY",
      "racy": "POSSIBLE"
    }
  }]
}

Frequently asked questions

Other comparisons