Type III Solar Radio Bursts Recognition Based on Mask R-CNN with Attention Mechanism

The high-energy signals generated by solar radio bursts can seriously affect the Earth's magnetic field, thereby affecting normal human production and life. By recognizing solar radio bursts, we can predict solar activities and prevent the serious impact of adverse space weather. This paper proposes to use Mask R-CNN to recognize Type III radio bursts, and improves traditional Mask R-CNN structure by adding CBAM attention mechanism, to improve the recognition performance of Mask R-CNN, achieving an accuracy of 91.3%.