Clothes classifying method based on convolutional neural network

Abstract

The invention discloses a clothes classifying method based on a convolutional neural network. The method comprises the following steps of acquiring clothes image samples, and dividing the samples into training samples and testing samples; preprocessing the training samples and the testing samples; constructing a convolutional neural network model; performing training of two stages including a forward propagation stage and a backward propagation stage on the convolutional neural network model through preprocessed training samples, finishing the training when the error calculated during the backward propagation stage reaches a desired value, and acquiring a parameter of the convolutional neural network model; testing the preprocessed testing samples by using the trained convolutional neural network model and outputting final clothes classifying results. The convolutional neural network model can make clothes images directly serve as network inputs, extract image features in an implicit way and establish a global feature expression. Compared with a manually designed feature extraction way, the method is more convenient and accurate. The problem that a conventional algorithm leads to low clothes classifying accuracy is solved.

Description

technical field [0001] The invention relates to a clothing classification method based on a convolutional neural network (ConvolutionalNeuralNetworks, CNN), which belongs to the technical field of image information processing. Background technique [0002] At present, researchers have proposed many algorithms for automatic clothing classification. Pan et al proposed to use BP neural network to identify knitted fabrics. Ben et al proposed a recognition method for knitted fabrics based on text features and support vector machines. Liu et al. proposed pose-based estimation and used features such as color, SIFT, HOG, etc. to classify clothes into 23 categories. Bourdev et al. developed a system to describe the appearance of people, using 9 attributes, such as male, T-shirt, long hair and other characteristics. In addition, the segmentation of clothes is also a research hotspot. proposed to use a constrained Delaunay triangle (CDT) based foreground and background estimation, ...

AI Technical Summary

Problems solved by technology

Due to the limitations of artificially designed features, the effect of traditional methods depends largely on whether the selected features are considered reasonable, which has a lot of blindness and generally has the problem of low classification accuracy.

Therefore, the current clothing classification algorithm has two main limitations:

First, traditional features cannot achieve satisfactory classification results, especially for categories with similar attributes

Second, there is currently no publicly available clothing database to objectively evaluate existing algorithms

Images