Driving scene target detection method based on deep learning and multi-layer feature fusion

Abstract

The invention relates to the technical field of traffic image processing, and discloses a driving scene target detection method based on deep learning and multi-layer feature fusion. The method comprises the following steps of 1) collecting a video image to serve as a training data set, and performing preprocessing; 2) building a training network; 3) initializing the training network to obtain a pre-training model; 4) performing training on the training data set by using the pre-training model obtained in the step 3), thereby obtaining a training model; 5) collecting a front image by using a vehicle-mounted camera, and inputting the image into the training model obtained in the step 4), thereby obtaining a detection result. The multi-layer feature fusion method based on a feature pyramid is adopted to enhance semantic information of a low-layer feature graph, so that the feature extraction quality of the network is improved, and higher detection precision is obtained.

Description

technical field [0001] The invention relates to the technical field of traffic image processing, in particular to a method for detecting objects in a driving scene. Background technique [0002] With the development of artificial intelligence technology, intelligent driving vehicles have achieved rapid development. Object detection is one of the key technologies for environment perception of intelligent driving vehicles. Accurate detection of traffic objects ahead helps to make correct driving decisions. In the target detection task in the driving scene, the road environment is complex and changeable, various targets block each other, and the illumination changes are complicated. These unfavorable factors restrict the improvement of the target detection effect. [0003] In the field of object detection, the ability to efficiently extract image features determines the quality of detection performance. Traditional machine learning methods use artificial features such as HOG ...

AI Technical Summary

Problems solved by technology

This method can achieve high detection accuracy, but the running speed is too slow, and the memory consumption is relatively large, which limits its application scenarios

The YOLO (You Look OnlyOnce) algorithm belongs to the first-level detection. It uses a 7×7 grid for end-to-end regression calculation, and combines the classification and positioning of the target. The detection speed of the YOLO algorithm is faster, but the simple grid regression operation makes Its detection accuracy is not high

The SSD (SingleShot MultiBox Detector) algorithm is also a representative of end-to-end detection. It introduces a multi-layer detection structure and improves the detection ability of multi-scale targets. However, the lack of semantic information of the low-level feature map makes the SSD algorithm weak in the detection of small targets.

Images