Skip to main content
The Image Engine verifies claims about images using deterministic methods first, with VLM fallback only for complex semantic claims.

​
Features

  • Metadata extraction — dimensions and format (PNG, JPEG, GIF, WebP)
  • Size verification — exact dimension comparison from metadata
  • Claim classification — routes claims to the appropriate verifier
  • Multi-VLM consensus — cross-validates semantic claims

​
Usage

from qwed_sdk import QWEDClient

client = QWEDClient(api_key="qwed_...")

# Read image
with open("chart.png", "rb") as f:
    image_bytes = f.read()

# Verify claim
result = client.verify_image(
    image=image_bytes,
    claim="The image is 800x600 pixels"
)

print(result.verdict)      # "SUPPORTED" or "REFUTED"
print(result.confidence)   # 1.0 (deterministic!)
print(result.methods_used) # ["metadata_extraction", "size_verification"]

​
Claim types

Claim TypeMethodDeterministic?
Size/DimensionsMetadata parsing✅ 100%
FormatHeader detection✅ 100%
ColorPixel sampling⚠️ Partial
TextOCR required❌ VLM
SemanticUnderstanding❌ VLM

​
Multi-VLM consensus

For semantic claims, use multiple VLMs: 
result = client.verify_image_consensus(
    image=image_bytes,
    claim="The person is smiling",
    min_agreement=2  # At least 2 VLMs must agree
)

print(result.agreement_count)  # 3/3
print(result.vlm_results)      # Individual VLM responses

​
Supported providers for image verification

The Image Engine supports multimodal verification through any provider that accepts image inputs. When you set ACTIVE_PROVIDER=gemini, QWED uses Google Gemini’s native multimodal capabilities for semantic image claims. Supported image formats are JPEG, PNG, and WebP. 
export ACTIVE_PROVIDER=gemini
export GOOGLE_API_KEY=your-google-api-key
See LLM configuration for full setup instructions.