The neural network source code and weights for target detection in radio astronomy
Peng Jia
This code implements a neural network designed for target detection in radio observation data. By leveraging the full three-dimensional structure of the data—comprising one frequency axis and two spatial axes—our method transcends traditional 2D stacking techniques. It utilizes 3D convolutional layers to simultaneously capture spatial morphology and spectral kinematic features. The network performs comprehensive 3D analysis to yield Bayesian inference results. These probability-based outputs provide not just point estimates but full posterior distributions for target parameters, which are crucial for downstream scientific study.
Files
Paper Information
Paper Title:
Automated Source Detection in Radio Astronomy Images Using Deep Neural Networks
Publication:
MONTHLY NOTICES OF THE ROYAL ASTRONOMICAL SOCIETY
Identifiers
CSTR:
11379.11.101756
DOI:
10.12149/101756
VO Identifier:
ivo://China-VO/paperdata/101756
Publication Date:
2026-06-23
Usage Statistics
Total Downloads
1739
Citations
Peng Jia et al. 2026. The neural network source code and weights for target detection in radio astronomy. Version 1.0. https://doi.org/10.12149/101756
@misc{10.12149/101756,
doi = {10.12149/101756},
url = {https://doi.org/10.12149/101756},
author = {Peng Jia},
title = {The neural network source code and weights for target detection in radio astronomy},
version = {1.0},
publisher = {Nataional Astronomical Data Center of China},
year= {2026}
}
doi = {10.12149/101756},
url = {https://doi.org/10.12149/101756},
author = {Peng Jia},
title = {The neural network source code and weights for target detection in radio astronomy},
version = {1.0},
publisher = {Nataional Astronomical Data Center of China},
year= {2026}
}
Versions
Main
This DOI represents all versions, and will always resolve to the latest one.
2026-06-23