Rotation robust three-dimensional model classification method based on feature attention

The invention discloses a rotation robust three-dimensional model classification method based on feature attention, and relates to the technical field of three-dimensional model classification, and the method comprises the steps: rendering a gray-scale map of a three-dimensional model, and obtaining a multi-angle view set of the three-dimensional model; designing a classification network to train the multi-angle view set to obtain a feature set; after the feature set is input into a feature attention module, attention operation is carried out on all features, and each attention obtains a weight, so that advanced features of the three-dimensional model are obtained; and designing an advanced feature classification network, and training the advanced feature classification network to obtain an advanced feature classifier for classifying the three-dimensional model. Feature attention is used to carry out differentiation processing on each feature, so that a subsequent network can process a view feature set having