About Accuracy & Limitations

AQI Vision uses atmospheric optics algorithms (Dark Channel Prior, edge density, and color statistics) to estimate PM2.5 and PM10 from camera images. This is a research-validated approach with the following caveats:

Expected accuracy: Category-level (Good / Moderate / Unhealthy etc.) is ~70–80% reliable outdoors. Exact µg/m³ values have ±30% uncertainty.

Best results when:
• Outdoors, pointing at open sky or distant scene
• Adequate daylight (not night-time)
• Phone held steady for 5+ seconds
• No glass/windows between camera and atmosphere

Not reliable for:
• Indoor measurements
• Pointing at walls or close objects
• Very dark or overexposed scenes
• Health-critical or regulatory decisions

For certified readings, use official monitoring stations or calibrated optical particle counters.