Bangla Image Captioning

A deep learning project for generating Bangla (Bengali) captions from images, built with PyTorch. This repository provides tools for training, evaluating, and using image captioning models on Bangla datasets.

Features

• End-to-end image captioning in Bangla
• Custom dataset support
• PyTorch-based modular code
• Easy training and evaluation scripts

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/Bangla-Image-Captioning.git
   cd Bangla-Image-Captioning

2. Install dependencies:

   • Python 3.6+
   • PyTorch 1.0 or later (pytorch.org)
   • NLTK

   Install with pip:

   pip install torch nltk

   Then, download NLTK data:

   import nltk
   nltk.download()

Dataset

• The default dataset is from BanglaLekha, but you can use your own dataset.
• Format: Each line in your CSV should be:

  /path/to/image1, "caption in Bangla"
  

  Example (train.csv):

  0001.jpg, "একজন মেয়ে হাত বাড়িয়ে বৃষ্টির পানি ধরার চেষ্টা করছেন "
  

Usage

• Training: Edit config.py as needed, then run:

  python train.py

• Vocabulary Building:

  python build_vocabulary.py

• Custom Dataset: Place your images and CSV in the appropriate folders and update paths in config.py.

Contributing

Contributions, issues, and feature requests are welcome! Feel free to open an issue or submit a pull request.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgements

• BanglaLekha for the dataset
• PyTorch and NLTK communities