Method and device for merging object detection information detected by each of object detectors

Abstract

The invention provides a method for merging object detection information detected by object detectors by using auto labeling and evaluation based on information exchange (V2X) between a vehicle and the outside. Each of the object detectors corresponds to each camera nearby, and detects objects in each of images generated from each of the cameras by image analysis based on deep learning. The methodincludes steps of: if first to n-th pieces of object detection information are respectively acquired from a first to an n-th object detectors in a descending order of degrees of detection reliabilities, a merging device generating (k-1)-th object merging information by merging (k-2)-th objects and k-th objects through matching operations, and re-projecting the (k-1)-th object merging informationonto an image, by increasing k from 3 to n. The method can be used for a collaborative driving or an HD map update through V2X-enabled applications, sensor fusion via multiple vehicles, and the like.

Description

technical field [0001] The present invention relates to a method and apparatus for self-driving vehicles, and more particularly, to a method and apparatus for integrating object detection information detected by object detection corresponding to each camera located in the perimeter . Background technique [0002] Generally, deep learning (Deep learning) is defined as a collection of machine learning algorithms that try to achieve high-level abstraction through a combination of multiple nonlinear transformation techniques, and is a machine learning that enables computers to learn the way people think in a large framework. a field of . [0003] Expressing some data in a computer-readable form, as an example, expressing the pixel information of an image in the form of a column vector (column vector), and conducting various researches to apply it to machine learning, through these efforts, such as Various deep learning technologies such as deep neural network, convolutional ne...

AI Technical Summary

Problems solved by technology

[0006] However, since existing object detectors detect objects based on learned parameters, there is a disadvantage that it cannot be confirmed whether objects are detected accurately in real situations, and separate monitoring is required for this

[0007] Also, it is difficult for existing object detectors to secure various learning data for performance improvement

[0008] Moreover, the existing object detector system can only confirm the direction that the camera can see and the area that is not blocked by obstacles, so the system has the problem of not being able to recognize the actual situation for the blind spot or the blocked area

[0009] Also, existing object detectors can output different results according to the learned parameters, so there can be discrepancies between the information detected by the surrounding systems

Images