RADiff : controllable diffusion models for radio astronomical maps generation

Abstract

Along with the nearing completion of the Square Kilometre Array (SKA), comes an increasing demand for accurate and reliable automated solutions to extract valuable information from the vast amount of data it will allow acquiring. Automated source finding is a particularly important task in this context, as it enables the detection and classification of astronomical objects. Deep-learning-based object detection and semantic segmentation models have proven to be suitable for this purpose. However, training such deep networks requires a high volume of labeled data, which is not trivial to obtain in the context of radio astronomy. Since data needs to be manually labelled by experts, this process is not scalable to large dataset sizes, limiting the possibilities of leveraging deep networks to address several tasks. In this work, we propose RADiff, a generative approach based on conditional diffusion models trained over an annotated radio dataset to generate synthetic images, containing radio sources of different morphologies, to augment existing datasets and reduce the problems caused by class imbalances. We also show that it is possible to generate fully synthetic imageannotation pairs to automatically augment any annotated dataset. We evaluate the effectiveness of this approach by training a semantic segmentation model on a real dataset augmented in two ways: 1) using synthetic images obtained from real masks, and 2) generating images from synthetic semantic masks. Finally, we also show how the model can be applied to populate background noise maps for simulating radio maps for Data Challenges. Code is available at: https://github.com/SKA-INAF/radiff Impact Statement—Deep learning methods have seen a wide application in several fields, including radio astronomy. A particular challenge in this domain is the difficulty of gathering annotated data to train supervised models for automatic source finding. In this work, we propose to address this limitation by proposing a generative model for data augmentation starting from handcrafted segmentation masks. Experiments show an improvement in performance of up to 18% when using real masks for generating images and 4% when augmenting with synthetic masks. Finally, we also employ the same model to populate largescale background noise maps with synthesized objects for radio map simulation in Data Challenges.