# Data requirements

To execute the code the following data are needed, once downloaded the path to the data must be specified in the misc/config.py file.

* [Ms-CoCo dataset (annotations and images)](http://cocodataset.org/#home)

* [Ms CoCo rest-val split](https://cs.stanford.edu/people/karpathy/deepimagesent/coco.zip)
from "Deep Visual-Semantic Alignments for Generating Image Descriptions" by Karpathy et al.

* [Word embedding](http://www.cs.toronto.edu/~rkiros/models/utable.npy) and [dictionnary](http://www.cs.toronto.edu/~rkiros/models/dictionary.txt) from the paper "Skip-Thought Vectors" by Kiros et al.

* [Pre-initialized weights of the image pipeline](https://cloud.lip6.fr/index.php/s/sEiwuVj7UXWwSjf)

## Additionnal data for localization evaluation

* [Visual Genome dataset (images and data and region descriptions)](https://visualgenome.org/)