Domestic traffic sign classification method based on Inception convolution module

Abstract

The invention discloses a domestic traffic sign classification method based on an Inception convolution module. The domestic traffic sign classification method comprises the following steps: constructing a domestic traffic sign data set; then, constructing a convolutional neural network based on an Inception convolution module; inputting the training set image data into a convolutional neural network, and performing training to obtain a model; and finally inputting the test set image data into the trained model to obtain a prediction result. The convolutional layer and the pooling layer are combined to replace a tiled layer and a full-link layer, and the simplified Inception convolution module is used, so that the number of parameters in the convolutional neural network can be effectively reduced, and the training frequency is reduced.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a domestic traffic sign classification method based on an Inception convolution module. Background technique [0002] In recent years, with the popularity of artificial intelligence, more and more scholars and experts have paid more and more attention to autonomous driving technology. How to accurately obtain road condition information necessary for autonomous driving has become a hot topic. Traffic signs are an integral part of road condition information, and the information they contain determines which mode the automated driving system should operate in. Traffic signs contain a lot of information, such as traffic restrictions, speed limits, height limits, warnings and forecasts of dangerous road conditions, etc., which provide an important auxiliary function for road safety. The recognition algorithms of traffic signs can be roughly divided into two categories: temp...

AI Technical Summary

Problems solved by technology

[0003] The last few layers of the traditional neural network are usually transmitted to the Softmax layer using the full connection layer. This method makes the last few layers of the full connection layer bring a lot of parameters when dealing with a network structure with more classifications, resulting in slower training speed.

Images