Skip to content

Latest commit

 

History

History
37 lines (20 loc) · 1.42 KB

README.md

File metadata and controls

37 lines (20 loc) · 1.42 KB
Raw

lartpc_minkowski

Deep learning based LArTPC event reconstruction using MinkowskiEngine sparse convolution APIs.

This repository is an extension/update of DeepLearnPhysics/lartpc_mlreco3d using the StanfordVL/MinkowskiEngine generalized spatio-temporal sparse convolution library.

This repository uses the generalized sparse convolution library provided by StanfordVL:
Github Link: MinkowskiEngine
Paper (arXiv link): 4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks

I. Contributing

  1. Please keep the 80 column limit, unless absolutely necessary.

II. Repository Structure

  • mlreco: contains all ML/DL algorithms for reconstructing LArTPC event.

  • iotools:

  • nn: shorthand for 'neural network'. This directory contains all neural network implementations that does not define a standalone model.

  • models: contains all standalone models that could be trained by calling bin/run.py.

  • post_processing: directory for post_processing algorithms, such as NMS and thresholding for PPN.

  • utils: miscellaneous utility functions.

  • bin:

  • test: contains unit tests.

  • config: configurations (*.cfg) files for training and validation.

III. Installation

We will use singularity containers to ship software.