Neural Network Training Methodology for Semantic Segmentation of Maps in Navigation Tasks

View at publisher Summarise with VIRAT
Authors: Vladimir V. Kovalev, Anton Salamakhin, Ekaterina V. Ilatovskaya, Anastasia O. Fedorkova, Denis M. Filatov
Cited by: 0
References: 1

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This paper presents a neural network training methodology for semantic segmentation of terrain maps, aimed at improving the accuracy of recognizing structural surface features such as buildings, roads, and natural objects. The proposed approach utilizes specialized convolutional neural network architectures and adaptive training strategies to achieve predictable segmentation quality depending on dataset characteristics. The developed method enables the generation of probabilistic class maps, enhancing the reliability of localization for autonomous systems in GPS-denied environments. The experimental results confirm the effectiveness of the proposed approach and demonstrate its potential for visual navigation and terrain analysis applications.

Fields of Study & Subjects

Advanced Neural Network Applications Robotics and Sensor-Based Localization Automated Road and Building Extraction

Publication details

Pages
70–75
Publication date
3 February 2026
Language
EN