ROAD INFRASTRUCTURE OBJECT RECOGNITION SOFTWARE PACKAGE
Authors:
1.
KNITU
Type:
Article
Pages:
from 153
to 157
Status:
In work
Language:
Russian
Keywords:
OBJECT RECOGNITION, SOFTWARE PACKAGE, MODULE, NEURAL NETWORK MODEL, RECOGNITION RESULT
Abstract (English):
The article presents the problem of recognizing road infrastructure objects based on the developed software package. The necessity of using YOLOv8 neural network models to solve this problem is substantiated. When implementing the software package, the following tools were selected: the Python programming language, the IDLE Shell 3.9.7 development environment. The software package was developed on a laptop with an Intel Pentium CPU 3825U processor, the frequency of which is 1.9 GHz, the amount of RAM is 8 GB, running a 64-bit version of the Windows 10 Pro operating system. The main elements of the software package are the detection module, the classification module and the experimental research module. The detection module includes a block of the road object recognition algorithm based on the YOLOv8 neural network. The classification module includes a block of object distribution by classes (road sign, road marking, traffic light), as well as a block of object recognition model accuracy assessment. The experimental research module includes object recognition on a test image or video file, obtaining the probability assessment result. The main window of the software package includes 3 tabs: «Model training» (contains tools for training YOLOv8 neural network models); «Object detection» (contains tools for loading images or video files, models for recognizing road infrastructure objects); «About the program» (contains information about the functionality of the software package). Examples of the software package functioning on a test image and a video file are given. In the future, the constructed neural network recognition models based on the YOLOv8 neural network can be used in a mobile application for automatic detection of objects (road markings, road sign, traffic light) in real time.
The article presents the problem of recognizing road infrastructure objects based on the developed software package. The necessity of using YOLOv8 neural network models to solve this problem is substantiated. When implementing the software package, the following tools were selected: the Python programming language, the IDLE Shell 3.9.7 development environment. The software package was developed on a laptop with an Intel Pentium CPU 3825U processor, the frequency of which is 1.9 GHz, the amount of RAM is 8 GB, running a 64-bit version of the Windows 10 Pro operating system. The main elements of the software package are the detection module, the classification module and the experimental research module. The detection module includes a block of the road object recognition algorithm based on the YOLOv8 neural network. The classification module includes a block of object distribution by classes (road sign, road marking, traffic light), as well as a block of object recognition model accuracy assessment. The experimental research module includes object recognition on a test image or video file, obtaining the probability assessment result. The main window of the software package includes 3 tabs: «Model training» (contains tools for training YOLOv8 neural network models); «Object detection» (contains tools for loading images or video files, models for recognizing road infrastructure objects); «About the program» (contains information about the functionality of the software package). Examples of the software package functioning on a test image and a video file are given. In the future, the constructed neural network recognition models based on the YOLOv8 neural network can be used in a mobile application for automatic detection of objects (road markings, road sign, traffic light) in real time.
Keywords:
OBJECT RECOGNITION, SOFTWARE PACKAGE, MODULE, NEURAL NETWORK MODEL, RECOGNITION RESULT
OBJECT RECOGNITION, SOFTWARE PACKAGE, MODULE, NEURAL NETWORK MODEL, RECOGNITION RESULT
