Skip to content

PyTorch-based Neural Networks for automating the finding of radio sources, components, and optical counterparts in International Low-Frequency Array (LOFAR) data

License

Notifications You must be signed in to change notification settings

jacobbieker/lofarnn

Repository files navigation

lofarnn

Second Master's Research Project for Leiden University, focused on attempting to use machine learning to identify radio sources and optical counterparts in LOFAR data

Installation

The easiest way to install this package is with pip with pip install lofarnn.

Otherwise, the lastest code can be built with pip install git+https://github.com//jacobbieker/lofarnn.git

Usage

The different PyTorch models and datasets can be easily imported from the lofarnn package. To preprocess LOFAR data into the correct format for either CNN or Detectron2 models, example code can be found under examples folder.

Models

PyTorch models used in the thesis are available here: https://drive.google.com/drive/folders/1lCFcQT7WRTiMxfd8jL2ReCoJrNAhj4BW?usp=sharing. The best performing model is the multi_cnn.pth model.

About

PyTorch-based Neural Networks for automating the finding of radio sources, components, and optical counterparts in International Low-Frequency Array (LOFAR) data

Resources

License

Stars

Watchers

Forks

Packages

No packages published