Acquaintance Recognition & Analysis

Problem

In home monitoring or office area management, a video or screenshot may need a quick answer: is this a known person, and where do they appear in the frame? Manually reviewing frames is slow and can miss multiple targets. Acquaintance Recognition & Analysis turns this into a structured task: enroll known faces first, then run face detection, feature extraction, and database matching against images or videos.

How It Works

The workflow is centered on the analysis script:
- Input: accepts local image/video paths or remote URLs, with the service downloading remote assets.
- Recognition: detects faces, extracts features, and compares them against the enrolled face database.
- Output: returns position labels, identity annotations, recommendations, and a report link for details.
- History lookup: uses --list to read the historical report list from the cloud API and filter by date range, avoiding local memory or manual aggregation.

Boundaries

This skill is useful for assistive acquaintance verification and area personnel recognition, not legal identity verification. Enroll target faces before recognition. Supported formats include jpg, jpeg, png, mp4, avi, and mov, with a maximum file size of 10MB. Historical reports should be treated as cloud-returned data, and results usually include position information and report links for follow-up.