Smart Feeder Pet Detection & Recognition

Problem

Smart feeders and IPC cameras generate fixed-angle video and image streams. Reviewing frames manually is slow, and multi-pet households need to know whether a specific cat or dog is present, not just whether a pet is present. This skill turns such media into structured findings about whether a cat or dog is detected, what it is, and whether it matches an enrolled individual.

How It Works

The skill accepts local files or public URLs, with media types such as video/image and common formats like mp4/avi/mov, jpg/png/jpeg. Key capabilities include:
- Pet detection and recognition: detects whether a cat or dog appears in the target area and returns structured monitoring results.
- Identity matching and enrollment: uses detect for recognition and enroll to register a pet ID/name into an individual profile database.
- Historical report lookup: fetches historical detection reports from the cloud API and renders them as a Markdown table, rather than relying on local memory or manual summaries.

Boundaries

It is best suited to fixed-view scenarios such as feeders and IPC cameras. Media should be clear and generally under 10MB. Results are useful for feeding management and care monitoring, but they do not replace human verification. Users do not supply identity parameters; the system handles association internally, and historical reports are read only from the cloud API.