Overview

The Sylvester ML API analyses a photo of a cat and estimates whether the cat shows signs of comfort or discomfort.

You send one image. The API:

  1. Detects the cat in the image and returns where it is (a bounding box) and how confident the detector is.
  2. Classifies the detected cat as comfort or discomfort with a probability score.

If no cat is found, or the detector isn't confident enough, the API tells you that instead of guessing.

Typical uses

  • Veterinary clinic intake and triage tools
  • Pet wellness and monitoring apps
  • Telehealth consultations
  • Insurance and shelter workflows that need a non-invasive signal
  • At home consumer care
  • At home pet hardware (e.g., smart litter boxes, feeders, cameras)

⚠️ The API gives a statistical pain signal to support human judgement. It is not a diagnosis. Sylvester.ai is not a diagnostic tool.

API versions at a glance

v2 (recommended) v1 (legacy)
Endpoint POST /v2/predict POST /invocations
Result labels comfort / discomfort Happy / Unhappy (plus mood_class 0 / 1)
Score Probability of discomfort (0–1), directly comparable to the threshold Confidence in whichever class was predicted
"No cat" / low quality HTTP 200 with a detection state HTTP 200 with success: false
Bad input Real HTTP errors (400 / 422) with a consistent error body Mostly HTTP 200 with success: false
Request tracing X-Request-ID header + request_id in every response None
Public, machine-readable docs Yes: OpenAPI at /v2/openapi.json No
Video No (send frames as images) Legacy endpoint only (deprecated)

New integrations should use v2. v1 remains available for existing customers.

Placeholders in code samples

  • https://api.example-sylvester-host: replace with the production base URL.
  • YOUR_API_KEY: replace with the key issued to your team.