HomoRAFT

This is the code of homography estimation model for my semester project at the ETH-Zurich. HomoRAFT is a deep learning model used to estimate homography between two images. The main structure is based on the RAFT.

Dependencies

If training with Synthetic MS-COCO dataset, please install the latest version of OpenCV, PyTorch and Kornia.

Training

To train the model, you need to first download the MS-COCO dataset, then change the config file to point to the correct path. Then, you can run the following command to train the model:

python train.py --config_file configs/raft-easy.yaml

Evaluation

To evaluate the model, first download the HPatch, then run the following command:

python evaluation/evaluate_hpatch.py --image-data-path <path-to-hpatch> --ckpt <path-to-model>

Weight can be downloaded from here