Infant Prone Sleep Suffocation Risk Detection

Problem

During infant sleep, prone positioning or mouth/nose occlusion by blankets, pillows, plush toys, or other objects can create suffocation risk, especially when adult supervision is delayed. Standard baby monitor footage often requires manual review and lacks a structured way to track posture, occlusion area, risk severity, and historical events.

How It Works

This skill processes video from a fixed camera above the crib and produces a risk alert report, not a medical diagnosis.

• Input: local mp4/avi/mov file path or public video URL, preferably under 10MB
• Detection: infant presence, sleep posture supine/side/prone/unknown, mouth/nose occlusion, and occlusion object such as blanket/pillow/plush_toy/bedding_fold/parent_arm
• Output: risk level low/medium/high/critical, risk_duration_sec, event_time, and snapshot_url
• History: use --list to query cloud-stored reports and render them as a Markdown table with report name, risk level, analysis time, and link

The open-id is mandatory and must be obtained from the user; historical reports must be fetched from the cloud API rather than local memory. Use the bundled script with parameters such as --input, --url, --open-id, and --detail.

Boundaries

It is suitable as an auxiliary safety module for smart baby cameras, cribs, or daycare monitoring, prompting caregivers to check the infant promptly. The output is visual analysis only and does not replace adult supervision, medical diagnosis, or emergency response. Infant video involves minor privacy; obtain caregiver consent and store recordings securely.