If you are looking for "how to get started" with hands-on experience, you are in the right place.
Below is a summary of tutorials made available voluntarily by our group members.
The most (if not all) of them are written in
python, and in fact in jupyter
Installing Jupyter Notebook
Jupyter notebook is not a requirement for using our software, but we share lots of examples in this format. So you might want to install it if you don't have one. Installation is same as many other python packages.
- If you don't have it, install either pip or conda python package manager.
- Either pip install or conda install
You might find a few glitches. Try googling a solution by copy-and-paste the error message (usually "package X not found") in the browser.
Additional packages: other extremely useful python packages include
Optional packages: want more? you can install
git clone our tutorial repository. All notebooks in this section can be found there (and you can execute/run the notebook by yourself as you follow the tutorial).
- Jupyter Notebook
- Introduction, very brief on our end. Features awesome Brilliantly wrong blog post by Alex Rogozhnikov)
- larcv Installation
- How to install
larcvsoftware (our C++ framework for file IO and image processing)
- How to install
- Opening larcv Data File
- What's in the file? Shows how to access images as numpy arrays and visualize them.
- Storing larcv Data File
- Brief demo on how to store your own data.
- Reading larcv Data File Fast
- Covers the basics of the tool you will be using for training a network.
These links are the blog posts that covered the contents of data files made available in our public data.
- Image classification sample
- Images of 5 different particle types, prepared for an image classification challenge.
- Semantic-segmentation sample
- Images of many particles prepared for semantic-segmentation challenge. Can be also used for object detection and instance-aware semantic segmentation algorithm training!
These are tutorial notebooks that actually train a toy algorithm for available open data. Like Quick Start, you can find these notebooks in our tutorial repository. But these examples use our public data which you have to download before executing the notebooks.
- Image classification training
- Train a network to classify an image containing one of 5 particles (e-, gamma, mu-, pi+, proton).
- Image classification inference
- Example of how to run an analysis (inference) using the trained network's weights.
- Semantic segmentation training
- Train U-ResNet for track/shower separation.
- Semantic segmentation inference
- coming soon
Quick Start serves as an introduction to LArCV. You can checkout the repository's wiki for more through guidance. Below you find some notebooks that are referred to in the wiki (but you can also just browse through here, if you want to!).
- Accessing file contents using IOManager
- LArCV's dedicated IO interface