smognet

• 

  Model Evaluation

• 

  Data analytics

• 

  Map

• 

  Home

SmogNet – Project Description

SmogNet is an end-to-end air quality intelligence system designed to monitor, analyze, and predict pollution patterns in major cities of Pakistan. It processes raw environmental sensor data, performs preprocessing and feature engineering, and detects abnormal pollution spikes using statistical and anomaly detection techniques such as Z-score, rolling IQR, and Isolation Forest.

After detecting spikes, the system classifies pollution sources (Crop Burning, Vehicular, Industrial, Dust Storm, or Mixed) using a hybrid approach that combines rule-based scoring with machine learning models like Random Forest and XGBoost trained on high-confidence pseudo-labeled data. Based on the detected severity and source, SmogNet generates structured public health alerts with recommendations for safety actions.

The project also includes interactive visualizations and a Streamlit-based real-time dashboard that allows users to input air quality values and receive instant predictions, risk levels, explanations, and charts. Overall, SmogNet provides a complete pipeline from raw data to actionable environmental insights for better air quality management and decision-making.

Built With

• css3
• flask
• frontend
• html5
• machine-learning
• python