A Hand-drawn Sketch Recognition Method Based on Attention Mechanism

Abstract

The invention discloses a hand-drawn sketch recognition method based on an attention mechanism. The method includes inputting the original hand-drawn sketch into a deep convolutional neural network to obtain a feature map output by the last convolution layer; inputting the feature map into a In the channel attention module, the optimized feature map based on channel attention is obtained; a classification network for predicting the vertical flip of hand-drawn sketches is trained, and the original hand-drawn sketches are input into the trained classification network to obtain a vertical flip spatial attention map; Combine the optimized feature map based on channel attention and the vertical flipped spatial attention map to calculate the feature map optimized by vertical flipped spatial attention; finally, output the recognition result through the fully connected layer. The advantages of the present invention are: channel attention and vertical flip space attention are used to optimize the features of the convolutional neural network, enabling the network to focus on learning more discriminative parts, thereby effectively improving the recognition accuracy of hand-drawn sketches.

Description

technical field [0001] The invention belongs to the field of computer vision, relates to hand-drawn sketch classification tasks, and in particular to a hand-drawn sketch recognition method based on an attention mechanism. Background technique [0002] Hand-drawn sketches can be regarded as an abstract form on a two-dimensional plane, which not only shows the information to be expressed, but also contains infinite space for imagination. It can be conveniently used to describe objects or scenes, outline storylines, design products or buildings, etc. It is widely used in painting and design work, such as comic production, urban planning and design, architectural composition, industrial design and fashion design, etc. . The related technologies of hand-drawn sketch recognition can be applied in many fields of computer vision, such as image retrieval and generation, 3D graphics retrieval and reconstruction, etc., so it has received more and more attention in recent years. [00...

AI Technical Summary

Problems solved by technology

However, due to the high abstraction of hand-drawn sketches, it is difficult to effectively model hand-drawn sketches, which makes the recognition accuracy of CNN and other deep network models on hand-drawn sketches drop sharply.

Images