metadata

title: Map-Data
emoji: 📊
colorFrom: green
colorTo: indigo
sdk: streamlit
sdk_version: 1.39.0
app_file: src/app.py
pinned: false
license: mit

High Fidelity Functional Map Generation for Industrial-Commercial-Residential Areas and Cluster Identification using Map Text Data Classification

Project Overview

This project aims to develop an automated system for generating high-fidelity functional maps of regions, classifying areas into industrial, commercial, and residential zones. It also identifies clusters within these classifications to optimize routes for Unmanned Aerial Vehicles (UAVs) conducting Air Quality Index (AQI) monitoring.

The system addresses the limitations of static AQI monitoring stations by leveraging machine learning, geospatial data, and UAV technology to create an efficient method for high-precision air quality monitoring across diverse urban and rural landscapes.

Project Structure

data/: Contains raw and processed data files
models/: Stores trained classification models
src/: Flask backend, Streamlit frontend code
utils/: Utility scripts for data loading and processing/classification rules
notebooks/: Jupyter notebooks for data preprocessing and classification
documentation/: Additional project documentation

Features

High-fidelity geospatial map generation
Region classification into residential, commercial, industrial, and other types
Data-driven route optimization for UAV-based AQI monitoring
Real-time AQI data collection across diverse regions

Data Processing

Data preprocessing steps are documented in the Jupyter notebooks located in the notebooks/ directory. These include:

Data collection from Maps API
Text data cleaning and normalization
Feature extraction, vectorization and clustering
Data labeling and augmentation

Models

Machine learning models used for text classification, including:

USE (Universal Sentence Encoder)
BERT (Bidirectional Encoder Representations from Transformers)
XGBoost
LSTM (Long Short-Term Memory networks)
Llama 3.2