Image-Captioning

This repository contains the code and models for image captioning using the COCO 2017 dataset.

Requirements

You can find the project dependencies in requirements.txt.

To download the COCO 2017 dataset, please follow the instructions on the official website.

Usage

To train a model, first, download the dataset and preprocess the images and captions using the preprocess.py script. The script can be run as follows:

python preprocess.py --data-root /path/to/coco2017 --output /path/to/output

This will create an HDF5 file containing the preprocessed images and captions.

Next, run the train.py script to train the model:

python train.py --input /path/to/preprocessed/data.hdf5 --output /path/to/output

To generate captions for new images using a trained model, run the generate.py script:

python generate.py --model /path/to/trained/model.pth --image /path/to/image.jpg

References

• Plotting examples: [https://cocodataset.org/#download](https://github.com/cocodataset/cocoapi/blob/master/PythonAPI/pycocoDemo.ipynb)
• COCO 2017 dataset: https://cocodataset.org/#download

License

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