YOLO v3-based detection method for key objects of transportation junction

Abstract

The invention discloses a YOLO v3-based detection method for key objects of a transportation junction. An algorithm is designed on the basis of the idea of direct regression to achieve multi-scale detection and multi-label classification. A Darknet-53 network based on ResNet is designed as a feature extractor based on the defects of current target detection technology, so that the detection precision and speed of YOLO technology architecture are improved. Meanwhile, the defect that the YOLO technology architecture is not good at detecting small objects is improved. The Darknet-53 network considers both the network complexity and the detection accuracy, and compared with a common target detection feature extraction network VGG-16, the model operand is reduced. The latest progress in the field of artificial intelligence is introduced into main target detection in the transportation hub, the detection precision and the detection speed are good, and meanwhile the method has the potential of being expanded and applied to other fields.

Description

technical field [0001] The present invention relates to the field of image processing, in particular to a method for detecting key objects in transportation hubs based on YOLO v3. Background technique [0002] With the rapid development of society, various high and new technologies continue to emerge, promoting the development of artificial intelligence. Among them, in the field of image processing, the technology of object recognition is developing rapidly. Image-based object detection technology is widely used in various industries. For example, image recognition technology is required in the fields of unmanned driving, unmanned supermarkets, remote sensing image recognition, biomedical detection, military and public security criminal investigation. Especially in the field of transportation, object recognition technology is gradually replacing the original technology for the detection and identification of pedestrians, motor vehicles, and non-motor vehicles. [0003] At ...

AI Technical Summary

Problems solved by technology

However, since YOLO adopts the idea of ​​direct regression, it only identifies and processes each frame of image as an independent data source, and the results processed for each frame of image are often poor in continuity and consistency.

Images