SAS Vision

무엇을 하는가

SAS VISION is a convolutional neural network (CNN)-based computer vision model designed to identify previously undocumented astronomical objects such as AGNs, NGCs, and stellar sources using large volumes of XMM-Newton observational data. This system fine-tunes open source CNN through transfer learning using XMM-Newton raw FITS files and their corresponding SAS pipeline-processed outputs. It performs a two-stage process: Denoising & Enhancement (using DnCNN, AutoEncoders, U-Net/W-Net, or DDPMs to preprocess and enhance noisy X-ray data), and Object Id...