Vehicle orientation detection method and system based on multi-task convolutional neural network

Abstract

The invention relates to a vehicle orientation detection method and system based on a multi-task convolutional neural network. The vehicle orientation detection method based on the multi-task convolutional neural network comprises the following steps: establishing a mature multi-stage multi-task convolutional neural network model; acquiring a real-time RGB image of a front vehicle, and inputting the real-time RGB image of the front vehicle into the multi-stage multi-task convolutional neural network model to obtain a real-time vehicle position, a real-time vehicle type, a real-time wheel position and real-time wheel and ground intersection position information of the front vehicle; optimizing the real-time wheel position and the real-time wheel and ground intersection position information,and obtaining the real-time vehicle orientation angle of the front vehicle through the real-time wheel and ground intersection position information. According to the method and system, the problems that the 3D detection frame of the preceding vehicle cannot be trained through deep learning and TTC calculation is inaccurate due to lack of preceding vehicle orientation information when the TTC is calculated by utilizing the pixel width change of the vehicle can be solved.

Description

technical field [0001] The invention relates to the technical field of automatic driving, in particular to a vehicle orientation detection method and system based on a multi-task convolutional neural network. Background technique [0002] In ADAS (Advanced Driving Assistance System, Advanced Driving Assistance System) monocular vision, accurate detection of the vehicle ahead is an essential function of the ADAS system, and 2D (two-dimensional) front vehicle detection is no longer sufficient to further improve the intelligence of the ADAS system Therefore, it is necessary to study the 3D (three-dimensional) frame detection technology of the front vehicle in monocular vision. By calculating the front vehicle orientation, the vehicle can be detected from 2D to 3D detection, and the key points where the front vehicle wheels cross the ground can effectively calculate the front vehicle orientation. angle. In addition, when calculating the collision time (TTC, Time to Collision) o...

AI Technical Summary

Problems solved by technology

[0005] The present invention provides a vehicle orientation detection method and system based on a multi-task convolutional neural network, which solves the problem that the 3D detection frame of the preceding vehicle cannot be trained through deep learning, and the lack of information on the orientation of the preceding vehicle when calculating TTC by using the change of the vehicle pixel width makes TTC calculation inaccurate question

Images