3D shape image classification method of isovariant 3D convolutional network based on partial differential operator

Abstract

The invention discloses a 3D shape image classification method of an isovariant 3D convolutional network model based on a partial differential operator, and the method comprises the steps of: carrying out the parametric modeling of a convolution kernel by employing the partial differential operator, solving a 3D rotating group and the feature domain of each convolution layer, obtaining an isovariant convolution kernel, and building an isovariant 3D convolutional network model PDO-e3DCNN, wherein the input of the PDO-e3DCNN is a 3D shape, and the output of the PDO-e3DCNN is prediction classification of the 3D shape, and the PDO-e3DCNN is used for 3D shape classification and identification visual analysis. According to the method, picture data with direction features can be effectively processed, and a lower 3D shape image classification error rate can be achieved on a data set by using fewer parameters.

Description

technical field [0001] The invention belongs to the technical fields of pattern recognition, machine learning and artificial intelligence, and relates to a 3D shape classification method, in particular to a 3D shape image classification method based on an equivariant 3D convolution network model of a partial differential operator. Background technique [0002] Over the past few years, convolutional neural network (CNN) models have become the dominant machine learning method for image recognition tasks. Compared with the fully connected network, a significant advantage of using CNN to process images is that they are translationally equivariant: the feature map obtained by first translating the image and then passing through several convolutional layers is the same as first passing the original image through the convolutional layer and then Translating gives the same result. In other words, each layer maintains translational symmetry, i.e. equivariance. Likewise, equivarianc...

AI Technical Summary

Problems solved by technology

However, this method can only be used to deal with the irreducible representation of the continuous group SO(3), and cannot be used to deal with the discrete group

Some contemporaneous or subsequent works, such as Tensor FieldNetwork (TFN) and LieConv, can only deal with the continuous group SO(3), and cannot include the discrete group

[0006] To sum up, the current equivariant 3D models can only deal with specific 3D rotation groups and their corresponding group representations, but cannot cover all commonly used groups and group representations under a unified framework

Images