Vehicle type recognition method based on rapid R-CNN deep neural network

Abstract

The invention discloses a vehicle type recognition method based on a rapid R-CNN deep neural network, which mainly comprises unsupervised deep learning, a multilayer CNN (Convolutional Neural Network), a regional advice network, network sharing and a softmax classifier. The vehicle type recognition method realizes a framework for implementing end-to-end vehicle detection and recognition by using one rapid R-CNN network in a real sense, and is capable of carrying out quick vehicle sub-category recognition with high accuracy and robustness under the environment of being applicable to the shape diversity, the illumination variation diversity, the background diversity and the like of vehicle targets.

Description

technical field [0001] The invention relates to the application of computer technology, pattern recognition, artificial intelligence, applied mathematics and biological vision technology in the field of intelligent transportation, in particular to a vehicle identification method based on a fast R-CNN deep neural network. Background technique [0002] The core function of the intelligent transportation system is the accurate detection of passing vehicles and the correct identification of vehicle types. The current research on vehicle detection and classification technology mainly includes two important technologies: automatic vehicle identification and automatic vehicle classification. [0003] Automatic vehicle identification is carried out by mutual recognition between on-board equipment and ground base station equipment. This technology is mainly used in toll collection systems. It is widely used in some developed countries, such as the AE-PASS system in the United States,...

AI Technical Summary

Problems solved by technology

[0020] To sum up, there are still several thorny problems in the recognition of vehicle models using deep neural network technologies such as convolutional neural networks: 1) How to accurately segment the overall image of the vehicle under test from the complex background; 2) How to use as little label image data as possible to accurately obtain the characteristic data of the vehicle model; 3) How to identify which model, which color, and which year the car was produced on the basis of the recognition of the vehicle model category; 4 ) How to automatically obtain the characteristics of vehicle models through deep learning; 5) How to balance recognition accuracy and detection efficiency while minimizing training and learning time; 6) How to design a classifier that meets the needs of vehicle models The classification requirements of the sub-categories, and the need to re-train the entire network after the car's appearance is updated; 7) How to design a framework that uses a CNN network to achieve end-to-end vehicle detection and recognition in the true sense; 8) How to reduce the influence of weather conditions and increase the adaptability of the system

Images