installation.md

Installation

This code is based on ScanRefer and ScanQA. Please also refer to the ScanRefer and ScanQA setup.

• Clone this repository:

  git clone https://github.com/matthewdm0816/BridgeQA.git
  cd BridgeQA
  

• Install PyTorch: pytorch==1.12.1 torchvision==0.13.1 compatible with your CUDA version.

• Install the necessary packages with requirements.txt:

  pip install -r requirements.txt
  

• Compile the CUDA modules for the PointNet++ backbone:

  cd lib/pointnet2
  python setup.py install
  

• Download BLIP checkpoints for VQA and Image-Text Retrieval, and put them to ckpts path.

Note that this code has been tested with Python 3.9.7, pytorch 1.12.1, and CUDA 11.3 on Ubuntu 20.04.1.