Vehicle detection method and system based on SSD (Single Shot MultiBox Detector) and vehicle gesture classification

Abstract

The invention discloses a vehicle detection method and system based on a SSD (Single Shot MultiBox Detector) and vehicle gesture classification. The method comprises the following steps that: according to an angle between a vehicle head and a horizontal axis, dividing a vehicle gesture; adding a vehicle gesture classification task into an original SSD network model; combining a vehicle detection loss with a vehicle gesture classification task loss to form a multi-task loss; replacing the softmax loss of the original SSD model with a flocal loss; carrying out joint optimization on the vehicle gesture classification task and the vehicle detection task; training to obtain a detection model; and utilizing the detection model to carry out vehicle detection on a picture to be detected to realizemultiscale and multi-angle vehicle detection. By use of the method, a deep learning target detection SSD is used for vehicle detection, the vehicle gesture classification is used for assisting the task and the vehicle detection task in carrying out joint training, and the flocal loss is added to solve the problem of unbalanced vehicle samples so as to improve the accuracy and the stability of thesystem.

Description

technical field [0001] The invention belongs to the field of image processing and pattern classification, and more particularly relates to a vehicle detection method and system based on SSD and vehicle posture classification. Background technique [0002] With the rapid development of artificial intelligence and computer vision, intelligent transportation system has become the development direction of modern transportation system, and vehicle detection is an important part of intelligent transportation system. Vehicle detection is mainly divided into two parts. The first part is the real-time detection of vehicles based on video. The detection of moving objects based on video is an important research direction of computer vision research. The rapid and accurate separation of moving vehicle objects from images is the basis for further research. Vehicle tracking can obtain a lot of traffic information, which is widely used in traffic monitoring, traffic incident detection and...

AI Technical Summary

Problems solved by technology

[0004] In view of the above defects or improvement needs of the prior art, the present invention provides a vehicle detection method and system based on SSD and vehicle attitude classification, thus solving the technical problems of low accuracy and poor robustness in the existing methods

Images