Training method of parking space structure detection model based on elastic polygon

Abstract

The invention relates to the field of intelligent driving, in particular to a training method of a parking space structure detection model based on an elastic polygon. The parking space structure model training method provided by the invention comprises: a step 101, acquiring a road sample image which is marked with parking space area information; a step 102, inputting the road sample image into apre-established initial neural network model; and a step 103, training the neural network model by using the sample image to obtain a parking space structure detection model. According to the invention, the independent parking space area information and the information of the points forming the independent parking space area are fused to obtain the independent parking space area represented by the elastic polygon, and the parking operation is carried out based on the parking space structure detection model.

Description

technical field [0001] The invention relates to the field of intelligent driving, in particular to a training method for a parking space structure detection model. Background technique [0002] With the development of science and technology, emerging concepts such as autonomous driving and unmanned vehicles have emerged as the times require. The automatic parking system is an indispensable part of the automatic driving technology. The interest in the development of the automatic parking system in the automotive industry is gradually increasing, and the parking space detection is one of the core perception tasks in the automatic parking problem. Specifically, the vehicle's camera or ultrasonic sensors can perceive the current environment of the vehicle, and the parking space detection technology is used to process the surrounding environment information of the vehicle acquired by the sensor to obtain nearby parking spaces, thereby guiding the parking of the vehicle. [0003]...

AI Technical Summary

Problems solved by technology

In fact, unfavorable factors such as ground materials, changes in lighting conditions, and shadows from nearby objects also increase the difficulty of vision-based parking space detection methods.

[0004] Existing parking space detection methods, one class usually uses the underlying edge and corner features to extract the parking space frame based on artificial construction rules, its feature expression ability is weak, and the rule-based method is difficult to extend to a variety of parking space forms

Another type of method is based on target detection technology, which uses a box to detect and extract parking spaces. This type of method is difficult to deal with oblique parking spaces, and the positioning of the parking space boundary box is also very imprecise.

There is another type of method to find the location of the parking space by integrating the detection information of the separation point and the separation line of the parking space. This method cannot directly provide the complete structural information of a parking space and it is difficult to give information about whether the parking space can be parked.

Images