Enhanced Textual Image Classification Using Ensemble Learning

Project Overview

This project explores the application of ensemble learning for textual image classification across multiple languages. By leveraging MobileNetV2 and ResNet50, it aims to enhance generalization, accuracy, and robustness. The ensemble model achieved a validation accuracy of 77%, outperforming individual models and demonstrating its potential in solving complex classification tasks.

---

Features

• Textual Image Classification: Classifies images based on textual content in multiple languages.
• Ensemble Learning: Combines MobileNetV2 and ResNet50 for improved performance.
• Efficient Training: Optimized architecture for faster training and inference.
• Multi-language Support: Works on textual images from 12 Indian languages, including Gujarati, Odia, Punjabi, Tamil, and others.

---

Dataset

The dataset includes images containing text from 12 Indian languages, collected and preprocessed for effective training and evaluation.

---

Setup and Installation

1. Clone the repository:

   git clone https://github.com/yourusername/repo-name.git

2. Navigate to the project directory:

   cd repo-name

3. Install dependencies:

   pip install -r requirements.txt
   

---

Model Architecture:

---

License:

This project is licensed under the MIT License.

---

Contact

For queries or collaboration, feel free to reach out: Email: [email protected]