Vehicle flow statistical method based on convolutional neural network vehicle model recognition in cloud environment

Abstract

The invention provides a vehicle flow statistical method based on convolutional neural network vehicle model recognition in a cloud environment. The vehicle flow statistical method comprises the steps of: constructing a vehicle model database; designing a convolutional neural network used for vehicle model recognition; training the convolutional neural network; carrying out vehicle model recognition by utilizing a real-time cloud platform; and tracking vehicles by utilizing a Kalman filter, and counting vehicle flow of corresponding vehicle models. According to the vehicle flow statistical method based on convolutional neural network vehicle model recognition in the cloud environment, an image recognition technology based on the convolutional neural network is combined with a cloud computing technology, vehicle flow statistics based on vehicle model recognition is subjected to parallelization by utilizing the cloud computing technology, and vehicle flow statistical rate is increased; moreover, on the premise of fully utilizing computing resources, an optimal network is found, the vehicle models are recognized, the vehicle flow is counted, and a real-time purpose is achieved.

Description

technical field [0001] The invention relates to the field of cloud computing, in particular to a traffic flow statistics method based on convolutional neural network model recognition in a cloud environment. Background technique [0002] In recent years, cloud storage and cloud computing technology have developed rapidly. The application of cloud computing has greatly improved the computing power and storage capacity of the system, reduced the difficulty of system deployment, and reduced management and maintenance costs. The development of cloud storage is conducive to promoting various video surveillance resources. Cloud integration effectively solves a series of problems encountered in the development of intelligent transportation systems such as massive video storage and image-based intelligent applications. [0003] Since deep learning is currently the mainstream of machine learning development, it is of great significance to try to apply deep learning to the traditional...

AI Technical Summary

Problems solved by technology

[0004] The existing vehicle type recognition has the problem that the high resolution of the image is contradictory to the recognition speed, and the algorithm is complex. At present, there is no vehicle flow statistics system based on deep learning vehicle type recognition.

Images