Our codes refer to https://github.com/yingxin-jia/SuperGlue-pytorch and https://github.com/magicleap/SuperGluePretrainedNetwork. We replaced the CNN backbone of encoder with equivariant steerable CNNs with the help of the e2cnn Pytorch package. SuperGlue is self-supervised training and does not require manual annotation.

Dependencies

• Python 3
• PyTorch >= 1.1
• OpenCV >= 3.4 (4.1.2.30 recommended for best GUI keyboard interaction, see this note)
• Matplotlib >= 3.1
• NumPy >= 1.18

Simply run the following command: pip3 install numpy opencv-python torch matplotlib

Or create a conda environment by conda install --name myenv --file superglue.txt

Contents

There are two main top-level scripts in this repo:

1. train.py : trains the superglue model.
2. load_data.py: reads images from files and creates pairs. It generates keypoints, descriptors and ground truth matches which will be used in training.

Download Data

Download the COCO2014 dataset files for training

wget http://images.cocodataset.org/zips/train2014.zip

Download the validation set

wget http://images.cocodataset.org/zips/val2014.zip

Download the test set

wget http://images.cocodataset.org/zips/test2014.zip

Training Directions

To train the SuperGlue with custom parameters, run the following command:

python train.py

Additional useful command line parameters

• Use --epoch to set the number of epochs (default: 20).
• Use --train_path to set the path to the directory of training images.
• Use --eval_output_dir to set the path to the directory in which the visualizations is written (default: dump_match_pairs/).
• Use --show_keypoints to visualize the detected keypoints (default: False).
• Use --viz_extension to set the visualization file extension (default: png). Use pdf for highest-quality.