Skip to content

Latest commit

 

History

History
119 lines (79 loc) · 1.84 KB

setup.md

File metadata and controls

119 lines (79 loc) · 1.84 KB

Setup

Download Image Matching Dataset

Install Kaggle

pip install --user kaggle

Download Kaggle API key Create the dir using this command and move the key to the dir

mkdir ~/.kaggle

Move the key to the dir

mv kaggle.json ~/.kaggle

CLI command to download kaggle dataset

kaggle competitions download -c 'image-matching-challenge-2023'

Unzip the dataset

unzip image-matching-challenge-2023.zip

Refrences: Kaggle Setup

Install Models LightGlue, ALiKeD, and DINOv2

Install LightGlue

pip install git+https://github.com/cvg/LightGlue.git

Create the dir for the checkpoints

mkdir -p ~/.cache/torch/hub/checkpoints

Download the LightGlue model

kaggle models instances versions download oldufo/lightglue/pyTorch/aliked/1

Unzip the model

tar -xzf lightglue.tar.gz

Move the model to the checkpoints dir

cp aliked_lightglue.pth ~/.cache/torch/hub/checkpoints/aliked_lightglue_v0-1_arxiv.pth

Download the ALiKeD model

kaggle models instances versions download oldufo/aliked/pyTorch/aliked-n16/1

Unzip the model

tar -xzf aliked.tar.gz

Move the model to the checkpoints dir

cp aliked-n16.pth ~/.cache/torch/hub/checkpoints/aliked-n16.pth

Download the DINOv2 model

kaggle models instances versions download metaresearch/dinov2/pyTorch/base/1

Create the dir for the model

mkdir -p dinov2/pytorch/base/1

`` Unzip the model

tar -xzf dinov2.tar.gz -C dinov2/pytorch/base/1

Setup Script

Use get_data.sh to download the dataset

bash get_data.sh

Use setup.sh to install the models

bash setup.sh