Abstract
Code to organize, prepare, and segment multi-scan, larger-than-memory, volumetric dataset. Deep learning segmentation uses pytorch-lightning for code organization, logging and metrics, and multi-GPU parallelization of training and prediction. Also uses moani for volumetric data augmentation and model implementation (currently U-Net++). Training can be done slice-by-slice with 2D models trained on 2D patches either aligned with or perpendicular to the electrode plane, or with 3D models on 3D patches. Sparse labels can be used to minimize required hand-labeling -- less than 0.1% of the entire dataset was labeled. Also includes code for data preparation (alignment/cropping/patching).
- Developers:
-
Ushizima, Daniela [1] ; Perlmutter, David [1]
- Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA (United States)
- Release Date:
- 2024-02-08
- Project Type:
- Open Source, Publicly Available Repository
- Software Type:
- Scientific
- Licenses:
-
BSD 3-clause "New" or "Revised" License
- Sponsoring Org.:
-
USDOEPrimary Award/Contract Number:AC02-05CH11231
- Code ID:
- 138909
- Site Accession Number:
- 2023-043
- Research Org.:
- Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA (United States)
- Country of Origin:
- United States
Citation Formats
Ushizima, Daniela, and Perlmutter, David.
batteryNET v0.0.1.
Computer Software.
https://github.com/lbl-camera/batteryNET.
USDOE.
08 Feb. 2024.
Web.
doi:10.11578/dc.20240730.2.
Ushizima, Daniela, & Perlmutter, David.
(2024, February 08).
batteryNET v0.0.1.
[Computer software].
https://github.com/lbl-camera/batteryNET.
https://doi.org/10.11578/dc.20240730.2.
Ushizima, Daniela, and Perlmutter, David.
"batteryNET v0.0.1." Computer software.
February 08, 2024.
https://github.com/lbl-camera/batteryNET.
https://doi.org/10.11578/dc.20240730.2.
@misc{
doecode_138909,
title = {batteryNET v0.0.1},
author = {Ushizima, Daniela and Perlmutter, David},
abstractNote = {Code to organize, prepare, and segment multi-scan, larger-than-memory, volumetric dataset. Deep learning segmentation uses pytorch-lightning for code organization, logging and metrics, and multi-GPU parallelization of training and prediction. Also uses moani for volumetric data augmentation and model implementation (currently U-Net++). Training can be done slice-by-slice with 2D models trained on 2D patches either aligned with or perpendicular to the electrode plane, or with 3D models on 3D patches. Sparse labels can be used to minimize required hand-labeling -- less than 0.1% of the entire dataset was labeled. Also includes code for data preparation (alignment/cropping/patching).},
doi = {10.11578/dc.20240730.2},
url = {https://doi.org/10.11578/dc.20240730.2},
howpublished = {[Computer Software] \url{https://doi.org/10.11578/dc.20240730.2}},
year = {2024},
month = {feb}
}