Respiratory Symptom Smart Recognition Tool

Problem

Respiratory symptoms such as coughing, expectoration, and wheezing are often tracked by observation, making it hard to know how frequently episodes occur, how long they last, or when they cross an abnormal threshold. This tool converts video behavior signals into countable indicators for early anomaly alerts, not medical diagnosis.

How It Works

• Input handling: Supports local video files or network URLs, including mp4, avi, and mov, with the documented maximum size of 10MB.
• Symptom recognition: Uses computer vision and behavior recognition, focusing on chest movement, posture, and mouth actions to distinguish normal breathing from coughing, expectoration, or wheezing.
• Frequency tracking: Logs episode time, frequency, and duration to support trend-style report content.
• Report output: Produces structured analysis, risk prompts, consultation guidance, and report links. Historical report queries are required to use the cloud API and should be rendered as a Markdown table.

The normal flow starts with preparing a video input, then the script handles identity association and analysis, followed by reviewing detected symptoms, alerts, and the report link. For history, use --list instead of relying on local memory or manually assembled records.

Boundaries

The tool is for auxiliary monitoring and early warnings. It does not replace physician diagnosis, clinical testing, or emergency judgment. Input video should be clear and relevant; poor framing, occlusion, very short duration, or atypical movements can reduce reliability. For diagnosis or treatment decisions, follow medical advice.