nNPipe: A neural network pipeline for automated analysis of morphologically diverse catalyst systems - Resources
Creators
- 1. Department of Materials, University of Oxford
- 2. Rosalind Franklin Institute, Harwell Research Campus
- 3. School of Chemistry, The University of Edinburgh
- 4. School of Chemistry, Cardiff University
- 5. Electron Physical Sciences Imaging Centre, Diamond Light Source; Johnson Matthey Technology Centre, Sonning Common
- 6. Department of Materials, University of Oxford; Rosalind Franklin Institute, Harwell Research Campus
Description
This dataset comprises of resources required to replicate the results described in "nNPipe: A neural network pipeline for automated analysis of morphologically diverse catalyst systems". nNPipe is a deep learning based method in which two deep convolutional neural networks are used for the automated analysis of 2048x2048 HRTEM images.
The file contains:
- Relevant experimental images as well as ground truth for Pd/C and Au/Ge systems.
- A workflow file explaining the nNPipe workflow.
- Mathematica 12.1 code for the generation of computational models.
- MATLAB code for HRTEM multislice simulations using MULTEM, as well as code required to form respective training datasets.
- Weights and files required for training the YOLOv5x module.
- Weights and files required for training the SegNet module.
- Mathematica 12.1 code required for reconstruction of 2048x2048 binary segmented maps of HRTEM images.
Files
nNPipe_resources.zip
Files
(1.3 GB)
Name | Size | Download all |
---|---|---|
md5:3a4cb8fe19fd961e837738ded62a0e2e
|
1.3 GB | Preview Download |
Additional details
Funding
- High temporal resolution TEM imaging of dynamic processes in heterogenous catalysts 2113841
- UK Research and Innovation
- Rosalind Franklin Institute Correlated Imaging Pump Priming EP/S001999/1
- UK Research and Innovation
- Rosalind Franklin Institute Correlated Imaging Phase 3 EP/T033452/1
- UK Research and Innovation