Multi-scale convolution method and device, terminal equipment and storage medium

• Main Authors: WANG SHUPENG, LIU YANGXING
• Format: Patent
• Language: Chinese; English
• Published: 21.06.2022
• 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 is suitable for the technical field of convolution processing, and provides a multi-scale convolution method and device, terminal equipment and a storage medium, and the method comprises the steps: inputting a to-be-processed matrix into a convolutional neural network according to a preset channel, the convolutional neural network comprising a convolution group corresponding to the preset channel; and performing multi-scale convolution calculation on the to-be-processed matrix through the convolution group, and outputting a target feature matrix. According to the scheme, the ability of the convolutional neural network to extract the multi-scale features of the to-be-processed matrix is improved through multiple groups of convolution. Convolutional layers of different scales do not need to be additionally added, so that the model processing effect can be improved on the premise of not increasing the calculation amount. 本申请适用于卷积处理的技术领域，提供了一种多尺度卷积的方法、装置、终端设备及存储介质，方法包括：将待处理矩阵按预设通道输入卷积神经网络，卷积神经网络包括与预