HAT-Match

HAT-Match: Graph Transformer with Hybrid Attention for Two-View Correspondence Pruning

PDF link

ECAI 2025

Requirements

Please use Python 3.7 and Pytorch 1.13.

Other dependencies should be easily installed through pip or conda.

pip install -r core/requirements.txt

Train

Train model on outdoor (yfcc100m) scene

python main.py --data_tr=yfcc-sift-2000-train.hdf5 --data_va=data_dump/yfcc-sift-2000-val.hdf5  --log_base=../model/yfcc_sift --gpu_id=0 

Train model on indoor (sun3d) scene

python main.py --data_tr=sun3d-sift-2000-train.hdf5 --data_va=sun3d-sift-2000-val.hdf5  --log_base=../model/sun3d_sift --gpu_id=0 

Test

Test pretrained model on outdoor (yfcc100m) scene

python main.py --run_mode=test --data_te=yfcc-sift-2000-test.hdf5  --model_path=../model/yfcc_sift/train/ --res_path=../model/yfcc_sift/test/ --gpu_id=0 

Test pretrained models on indoor (sun3d) scene

python main.py --run_mode=test --data_te=sun3d-sift-2000-test.hdf5  --model_path=../model/sun3d_sift/train/ --res_path=../model/sun3d_sift/test/ --gpu_id=0

Results

-/- : without/with RANSAC

Acknowledgement

OANet/CLNet/BCLNet, etc.