Driverless environment target detection method based on embedded equipment

Abstract

The invention discloses a driverless environment target detection method based on embedded equipment. The method comprises the steps of firstly completing the recognition of a target through employing an improved YOLOv3 algorithm, carrying out the improvement of a dimension clustering algorithm K-Means of the YOLOv3, obtaining an optimal anchor frame scale adaptive to a data set through employing an improved K-Means + + algorithm, and improving the clustering precision; secondly, for a Darknet-53 backbone network in YOLOv3, shortening the forward reasoning time of the model by combining a BN layer and a convolutional layer of the backbone network; and finally, realizing great reduction of model parameters through a model compression technology so as to be applied to embedded equipment, analyzing real-time video data input by a camera, and judging a target category in a current scene.

Description

technical field [0001] The method relates to an unmanned driving environment target detection method based on an embedded device, and belongs to the technical field of unmanned vehicle environment perception. Background technique [0002] In the early stage, there were some methods for image target detection, the core invention points of which were obviously different from the present invention, and the following technical solutions were relatively close after retrieval. [0003] Application number: CN201911153078.6 provides a target detection system suitable for embedded devices, including embedded devices and servers; the target detection logic running on embedded devices consists of multi-layer shared basic network, private basic network and detection module Composition; the parameters of the shared basic network come directly from the output of the upper layer; the image is processed by the shared basic network and the private basic network to obtain the feature map, and...

AI Technical Summary

Problems solved by technology

2) The technical solutions are different. The image features of the above method are fusion features, which are composed of RGB image and depth image features, but the image features used in this method do not use depth features.

Images