[Deep Learning] - 2023/2024

Authors:

• Iván Hernández, [iv97n]
• Marcel Fernández, [u198734]
• Alejandro Pastor, [u199327]

Dataset: COCO Dataset, Image Captioning 2014

TL;DR

This repository contains code for the execution of a web-based image caption generator, which has been designed based on the architecture described in the paper Show and Tell: A Neural Image Caption Generator.

Presequisites

• Python Poetry

Minimal setup

• Clone the repository into your local machine

git clone git@github.com:iv97n/image-captioning.git

• In the root directory of the project create a folder called model for storing the pretrained model .ckpt file

cd image-captioning
mkdir model

• Download the nic.ckpt file containing the model weights and save it into the model directory you just created
• Also in the root directory create a folder called data for storing the images you would like to pass to the model

mkdir data

• Install the dependecies running poetry install in the project root directory. This might take a few minutes.

poetry install

• Now you are ready to rock some captions 😎

Caption generation

• Make sure the image you want to caption is within the data folder
• Run the following command, substituting example.jpg with your image name

poetry run python main.py --imagepath data/example.jpg