Deep learning training model
This folder includes the Deep-learning neural network model used to segment the µCT volumes and an excel spreadsheet with the hyper-parameters used for training. The data provided are (i) a Deep Learning Model (DLM) for micro-Computed X-ray tomographic Images. The DLM is specifically created for seg...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Software |
| Language: | English |
| Published: |
University of Edinburgh. School of GeoScience
2023
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10283/8508 https://doi.org/10.7488/ds/7493 |
| Summary: | This folder includes the Deep-learning neural network model used to segment the µCT volumes and an excel spreadsheet with the hyper-parameters used for training. The data provided are (i) a Deep Learning Model (DLM) for micro-Computed X-ray tomographic Images. The DLM is specifically created for segmenting mineral phases within the volumetric data for the Gypsum to Bassanite dehydration reaction. The DLM was created with the Software ORS Dragonfly and can be opened only within that software; (ii) the .jason metadata related to the DLM, providing all the hyper-parameters used for training the DLM and the Outputs of the loss function tracking the performances of the training. (iii) A log file (.csv) containing the Loss function values produced during training and used to assess the quality and the accuracy of the trained DML. (iv) a .jason file containing the information relative to the hyper-parameters (e.g., the type of Convolution Neural Network architecture used to create the DLM, the Batch, Stride Ratio and Patch size of the used during training). The dataset relates to the upcoming publication "Using Internal Standards in Time-resolved X-ray Micro-computed Tomography to Quantify Grain-scale Developments in Solid State Mineral Reactions" by Roberto Emanuele Rizzo, Damien Freitas, James Gilgannon, Sohan Seth, Ian B. Butler, Gina McGill, Florian Fusseis. |
|---|