Typhoon intensity determination method and system based on point cloud voxels, equipment and medium

• Main Author: TAN JINKAI
• Format: Patent
• Language: Chinese; English
• Published: 15.02.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention relates to the technical field of deep learning, in particular to a typhoon intensity determination method and system based on point cloud voxels, equipment and a medium. The method comprises the following steps: carrying out deep convolution and fusion on point cloud voxel features through different perception domains, and acquiring fusion features; performing channel shuffling and expansion on the fusion feature to obtain a first point cloud feature; performing feature extraction on the first point cloud feature by using a channel attention mechanism to obtain a second point cloud feature; inputting the second point cloud feature into a three-dimensional sparse convolutional layer for sparsification to obtain a third point cloud feature; inputting the third point cloud feature into a three-dimensional compression convolution layer for feature compression to obtain a fourth point cloud feature; and acquiring typhoon intensity through prediction according to the fourth point cloud feature. Accor