Totally Defined Nanocatalysis

The repository contains the code for the article "Towards Totally Defined Nanocatalysis: Deep Learning Reveals the Extraordinary Activity of Single Pd/C Particles" by Dmitry B. Eremin, Alexey S. Galushko, Daniil A. Boiko, Evgeniy O. Pentsak, Igor V. Chistyakov, and Valentine P. Ananikov

• tdnc/ — a Python module, which contains useful functions and classes for training segmentation models for SEM image analysis
• configs/ — task-specific configuration files, allowing to reproduce the models
• train.py — the segmentation model training script
• inference.py – the code to perform inference of the models
• configs/augmentations/ – augmentations, specific for each material

All models can be retrained with the following command

for f in configs/*.yaml; do python train.py --config $f; done

The training was performed on a single NVIDIA 1080 TI.