BUILDING SMOKE DETECTION POWERED BY MACHINE LEARNING & DEEP LEARNING
Real-time IoT sensor analysis using 7 ML models plus a MobileNetV2 CNN for image-based fire/smoke detection. Sub-200ms prediction from temperature, humidity, CO₂, TVOC & particulate sensor data.
How SmokeGuard Works
SENSOR INGESTION
IoT sensors capture temperature, humidity, CO₂ (eCO₂), TVOC, raw H₂ & ethanol, pressure, and PM1.0/PM2.5 particulate readings in real time.
PREPROCESSING
Data is normalized via a fitted scaler ensuring consistent model input regardless of sensor drift or environmental variation across devices.
ML CLASSIFICATION
7 trained classifiers including Random Forest, SVM, Gradient Boosting analyze features to output a binary smoke/no-smoke decision.
INSTANT ALERT
Results are returned with confidence scores. Positive detections trigger immediate visual alerts and logging for building management.
7 Trained ML Models + 1 CNN
Zero-Latency Threat Classification
When sensors detect anomalous readings, SmokeGuard cross-references data patterns across all 12 input features to deliver a definitive threat assessment in under 200ms.
Ready to Protect Your Building?
Deploy SmokeGuard AI for instant ML-powered smoke detection.