INSTALL.md

Installation Instructions

Install Dependencies

1. Clone the repository:

   git clone git@github.com:arjunmajum/vln-bert.git
   

2. Create a conda environment:

   cd vln-bert
   conda create -n vlnbert python=3.6
   conda activate vlnbert
   

3. Install additional requirements:

   pip install -r requirements.txt
   

4. Install NVIDIA Apex by following their quick start instructions.

Data Preprocessing

1. The code in this repository expects a variety of configuration and data files to exist in the data directory. The easiest way to get all of the required configuration files is to run the following command:

   python scripts/download-auxiliary-data.py
   

2. Next, precompute image features using bottom-up-attention (i.e., Faster R-CNN pretrained on Visual Genome) from panos in the Matterport3D dataset. Follow the steps outlined in scripts/matterport3D-updown-features/README.md.

   Alternatively, you can download and extract the the data from here:

   matterport-ResNet-101-faster-rcnn-genome.lmdb [13.8 GB]

Verify the directory structure

After following the steps above the data directory should look like this:

data/
  beamsearch/
  config/
  connectivity/
  distances/
  logs/
  matterport-ResNet-101-faster-rcnn-genome.lmdb
  models/
  runs/
  task/