Skip to main content

​
Overview

The CompareResponse model represents the output from the face comparison endpoint. It contains an array of match results with probability scores for each person compared against the target face.

​
Rust Definition

#[derive(Serialize)]
pub struct CompareResponse {
    pub matches: Vec<MatchResult>,
}

​
TypeScript Definition

interface CompareResponse {
  matches: MatchResult[];
}

​
Fields

​
matches
MatchResult[]
required
An array of match results containing the name and probability score for each person compared. The array is typically sorted by probability in descending order, with the highest matches first.

​
Example JSON

{
  "matches": [
    {
      "name": "Jane Smith",
      "probability": 0.9823
    },
    {
      "name": "John Doe",
      "probability": 0.4521
    },
    {
      "name": "Bob Johnson",
      "probability": 0.2134
    }
  ]
}

​
Response Interpretation

​
Probability Scores

The probability values range from 0.0 to 1.0:
  • 0.9 - 1.0: Very high confidence match - likely the same person
  • 0.7 - 0.9: Strong match - high probability of being the same person
  • 0.5 - 0.7: Moderate match - some similarity but less certain
  • 0.0 - 0.5: Low match - likely different people

​
Interpreting Results

The matches array includes all people from the request, each with their computed similarity score. Higher probability values indicate stronger facial similarity to the target image. For production use cases, consider:
  • Setting a probability threshold (e.g., 0.7) for positive matches
  • Using the highest probability match when identifying a single person
  • Accounting for image quality, lighting, and angle variations

Build docs developers (and LLMs) love

Get started for freeTalk to us