Unattended Care Monitoring

Problem addressed

When an elderly person lives alone, it is hard to tell from a single phone call or check-in video whether the home has had no visitor interaction for a long time. This skill is aimed at fixed-camera surveillance material and helps family members detect an unattended state, then create a traceable care reminder when a threshold such as 24 hours is exceeded.

How it works

• Core capabilities: person detection, interaction behavior recognition, visitor count, unattended duration tracking, and alert triggering.
• Input and output: accepts local files or network URLs in common formats such as jpg/jpeg/png/mp4/avi/mov; returns a structured report, monitoring results, risk notes, and recommendations.
• Standard flow: start with clear surveillance material, run a single analysis, review the risks and recommendations in the report, and query historical reports when needed.
• History lookup: report lists are retrieved from the cloud API rather than assembled from local memory, temporary summaries, or long-term cache, making the output easier to audit.
• Boundaries: accuracy depends on camera angle and footage clarity; the result is for care reference only and does not replace in-person visits, emergency response, or professional nursing.