METHOD AND APPARATUS FOR PERFORMING CONVOLUTION NEURAL NETWORK OPERATIONS USING 3X3 CONVOLUTION MATRIX

    公开(公告)号:WO2022197325A1

    公开(公告)日:2022-09-22

    申请号:PCT/US2021/053281

    申请日:2021-10-03

    Abstract: A method and apparatus for performing a convolution of a NxN matrix. A weights matrix for a NxN Convolutional Neural Network (CNN) is received and is divided into 3x3 weights matrixes. Lines of image values are read and are stored in a buffer as sets of image values. A 3x3 convolution is performed to generate a 3x3 convolution value. All 3x3 convolution values associated with a particular NxN convolution and a particular set of image values are summed. The 3x3 convolutions and the summing are repeated until all columns in the set of image values have been processed; and the reading, the storing, the performing 3x3 convolutions, the summing and the repeating performing 3x3 convolutions are repeated until all lines of image values have been processed. The sums associated with a particular NxN convolution are added together to generate an NxN convolution value for each of the NxN convolutions.

Patent Agency Ranking