The best way to conference proceedings by Francis Academic Press

Web of Proceedings - Francis Academic Press
Web of Proceedings - Francis Academic Press

Pedestrian Detection Method Based on Neural Network and Data Fusion

Download as PDF

DOI: 10.25236/cseem.2020.184


Juan Fu

Corresponding Author

Juan Fu


Pedestrian detection is a research topic that has attracted much attention in computer vision, and it has a wide range of applications in many occasions such as public safety, scene monitoring, and traffic operations. The purpose of this article is to use neural network and data fusion technology to detect pedestrians and improve traffic safety. This article first analyzes the pedestrian detection technology based on neural network and data fusion, and summarizes the basic knowledge of convolutional network applications from four aspects: network structure design, loss function design, regularization method and optimization strategy; Take handwritten digit recognition as an example. It focuses on comparing the differences between different variants of gradient descent optimization algorithms. It uses many of the most popular algorithms, applies graph theory and ID3 algorithms, and uses post-pruning techniques to implement classification decision trees. The classification rules are generated, and the application scenarios under the fusion of neural networks and data are completed. The classification decision tree model is constructed. Secondly, based on the leading methods in the target detection field, based on the summarized design criteria and the scale characteristics of pedestrians, adjusted The anchor window setting of the network and the area generation network method have added the environment area pooling layer. Then based on the open source deep learning framework, the network is implemented on the pedestrian data set. The experimental results show that this method can achieve efficient pedestrian detection. According to the statistical results, the detection efficiency has reached 90%.


Pedestrian Detection; Application of Graph Theory and ID3 Algorithm; Classification Decision Tree Model