55-alt/tunisian-architectural-classification

Multimodal AI project for classifying Tunisian architectural identity using tabular data, floor plans and building facades. \# Multimodal AI for Tunisian Architectural Identity Classification \## 📌 Project Overview This project explores the use of Artificial Intelligence and Machine Learning to analyze and classify Tunisian architectural identity. The main objective is to investigate whether combining different sources of architectural information can help classify houses into three categories: \- Traditional \- Mixed \- Modern The project follows a multimodal approach by combining structured architectural data with visual information extracted from floor plans and building facades. \--- \## 🎯 Objectives The main objectives of this project are: \- Analyze architectural characteristics of Tunisian houses. \- Classify houses according to their architectural style. \- Extract visual features from floor plans and facades. \- Compare different types of input data. \- Investigate the contribution of visual information to classification. \- Explore multimodal feature fusion. \- Evaluate the performance of different machine learning approaches. \--- \## 🧠 Methodology The project follows several main stages: 1\. Data preprocessing and exploration 2\. Analysis of architectural tabular data 3\. Image preprocessing 4\. Visual feature extraction using VGG16 5\. Dimensionality reduction using PCA 6\. Unsupervised exploration using K-Means 7\. Classification using Random Forest 8\. Comparison of different data modalities 9\. Multimodal feature fusion 10\. Model evaluation \--- \## 📊 Data Modalities The project uses three main sources of information: \### Tabular Architectural Data Structured information describing architectural characteristics of the studied houses. \### Floor Plans Visual information extracted from architectural floor plans. \### Building Facades Visual information extracted from building facade images. The different modalities are evaluated separately and in combination to understand their con …