Skip to content
A modular repository for developing Image Captioning Approaches
Jupyter Notebook Other
Branch: master
Clone or download

Latest commit

Fetching latest commit…
Cannot retrieve the latest commit at this time.

Files

Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
analysis
cnn
configs
datasets
language
models
preprocessing
utils
.gitignore
LICENSE
README.md
requirements.txt

README.md

Image Captioning Approaches

Build and experiment scalable Image Captioning techniques.

Dataset Loaders

Data preparation and loading is complete for the following Image captioning datasets

  1. Flickr8k
  2. Google Conceptual Captions

Download Conceptual Captions

  1. The main dataset is available for download at https://ai.google.com/research/ConceptualCaptions/download

Steps to download images:

  1. Download the training split from the dataset page
  2. Set the GOOGLE_CAPTIONS_FILE url path in preprocessing\GoogleCaptions.py.
  3. Choose the number of samples to download by setting NUM_SAMPLES
  4. Run the script using python3 GoocleCaptions.py
  5. The images would be downloaded in the set directory

Keras Loaders

  1. The keras data loaders are written for Flickr8k and Google Conceptual Captions at datasets\flickr8k and datasets\googlecc.py.
  2. Currently, there is no separate usage of Validation set and a part of the training data is used for Validation.

Training

You can’t perform that action at this time.