A road scene semantic segmentation method based on convolutional neural network

Abstract

The invention discloses a road scene semantic segmentation method based on a convolutional neural network. In the training stage, a convolutional neural network is constructed, and its hidden layer includes 5 neural network blocks, 5 transitional convolutional layers, and 5 jumps. Deconvolution block, 4 cascade layers; use the original road scene image input to the convolutional neural network for training, and obtain the corresponding 12 semantic segmentation prediction maps; and then calculate the 12 semantic segmentation images corresponding to the original road scene image The loss function value between the set of segmentation prediction images and the set of 12 one-hot encoded images processed by the corresponding real semantic segmentation images, to obtain the optimal weight vector and bias item of the convolutional neural network classification training model ; In the test phase, the road scene image to be semantically segmented is input into the convolutional neural network classification training model to obtain a predicted semantically segmented image; the advantage is that the semantic segmentation efficiency and accuracy of the road scene image are improved.

Description

technical field [0001] The invention relates to a deep learning semantic segmentation method, in particular to a convolutional neural network-based semantic segmentation method for road scenes. Background technique [0002] The rise of the intelligent transportation industry has led to more and more applications of semantic segmentation in intelligent transportation systems. From traffic scene understanding and multi-target obstacle detection to visual navigation, semantic segmentation technology can be used to achieve. Currently, the most commonly used semantic segmentation methods include algorithms such as support vector machines and random forests. These algorithms mainly focus on binary classification tasks to detect and recognize specific objects such as road surfaces, vehicles, and pedestrians. These traditional machine learning methods often need to be implemented through high-complexity features, but it is simple and convenient to use deep learning to semantically ...

AI Technical Summary

Problems solved by technology

[0004] Most of the existing road scene semantic segmentation methods use deep learning methods. There are many models that combine convolutional layers and pooling layers. However, the feature maps obtained by purely using pooling operations and convolution operations are single and not representative. , which will lead to the reduction of the feature information of the obtained image, which will eventually lead to rough restoration of the effect information and low segmentation accuracy.

Images