Pytorch model only of U-Net, as used in the paper:
Pasquale Lafiosca et al., Automatic Segmentation of Aircraft Dents in Point Clouds, SAE Technical Paper, 2022, https://doi.org/10.4271/2022-01-0022
If you find this useful, please cite as
@inproceedings{dentNet,
title={Automatic segmentation of aircraft dents in point clouds},
author={Lafiosca, Pasquale and Fan, Ip-Shing and Avdelidis, Nicolas P},
year={2022},
institution={SAE Technical Paper}
doi={10.4271/2022-01-0022}
}
In the paper, the loss function used was nn.BCEWithLogitsLoss and the optimizer was optim.Adam.
The real samples mentioned in the paper and their predictions are shared in the realSamples folder.
The training dataset cannot be shared.
Please note that this repository is archived.