You can deploy this model and start earning money today!

Lhaippp/
DMHomo

Name: DMHomo
Rating: 2.5 (6 reviews)

Views

No views yet

DMHomo: Learning Homography with Diffusion Models

dl.acm.org/doi/10.1145/3652207

To-Do List

Release the pre-trained models and inference code for the HEM (Homography Estimator Module) and DGM (Data Generator Module).
Release the generated dataset based on the CA-Homo Dataset and the trained DGM.
Release the training scripts for HEM and DGM.

Inference Script for HEM

we use accelerate for multi-GPUs processing

# Please set the path of 'CA-Homo Dataset' by [test_data_dir] in HEM/experiments
accelerate launch hem_evaluate.py --model_dir HEM/experiments --restore_file HEM.pth -ow

Thanks

Our framework builds upon previous benchmarking works; we offer our gratitude to them, including, but not limited to:

Citation

If you use this code or ideas from our paper for your research, please cite our paper:

@article{li2024dmhomo,
  title={DMHomo: Learning Homography with Diffusion Models},
  author={Li, Haipeng and Jiang, Hai and Luo, Ao and Tan, Ping and Fan, Haoqiang and Zeng, Bing and Liu, Shuaicheng},
  journal={ACM Transactions on Graphics},
  year={2024},
  publisher={ACM New York, NY}
}

Lhaippp/DMHomo