| Name: Srinjana Routh |
| Affiliation: ARIES |
| Conference ID: ASI2026_250 |
| Title: A Detectron 2 based pipeline for instance segmentation of filaments across datasets |
| Abstract Type: Poster |
| Abstract Category: Sun, Solar System, Exoplanets, and Astrobiology |
| Author(s) and Co-Author(s) with Affiliation: Srinjana Routh(Aryabhatta Research Institute of Observational Sciences), Vaibhav Pant(Indian Institute of Technology, Delhi), Dipankar Banerjee(Indian Institute of Space Science and Technology, Thiruvananthapuram) |
| Abstract: We present SUITron, a deep learning framework for real-time detection and segmentation of solar features like filaments and active regions . Leveraging the robust architecture of Detectron2, an open-source object detection platform developed by Facebook AI Research, SUITron combines high-quality implementations of Mask R-CNN and related models to deliver state-of-the-art performance on solar data. This pipeline enables rapid identification of critical structures such as active region, plages and filaments to give realtime parameters including area and chirality of which are being integrated into further analyses for space weather prediction in real time. The utility of the model will be demonstrated using GONG datasets. The resulting system offers a scalable toolset for the Aditya-L1 Support Cell, where it is to be integrated in the near future and is expected to facilitate continuous, high-resolution monitoring of solar activity crucial to heliophysics and space mission operations.
|
