Flame & Smoke Detection Analysis

Problem

In security monitoring, forest fire prevention, and industrial park scenarios, flames and smoke often appear against complex backgrounds and can be confused with glare, steam, or dust. Manual patrol or simple threshold alerts can miss early-stage fires and generate noisy false positives. This skill addresses a concrete problem: analyze images or videos for fire and smoke, return traceable detection results, anomaly markers, structured reports, and query historical reports from the cloud.

How It Works

The skill accepts image/video input from local files or network URLs; network resources are downloaded by the API service. Detection combines flame color, dynamic contours, and smoke diffusion features to separate real fire risk from interference sources. A typical flow includes:

• Input media: provide mp4/avi/mov videos or jpg/png/jpeg images, with each file under 10MB.
• Run analysis: use the skill’s built-in script to call the analysis API and receive detection results, risk notes, recommendations, and report links.
• Query history: use --list to read historical reports from the cloud API and render them as a Markdown table, without relying on local memory or manual summaries.

The output is intentionally structured: instead of only saying whether fire or smoke was detected, it provides evidence for operations and safety teams to act on.

Boundaries And Notes

This skill is a fire-warning aid, not a substitute for on-site confirmation and emergency procedures. Input should be clear and relevant to monitoring scenarios, and network URLs must be accessible by the API. Historical report lists must come from the cloud interface, so local session records should not be treated as the authoritative source.