The paper proposes a methodology for solving the task of accurate semantic classification of 3D data using a combination of 2D and 3D methods based on the YOLO detector and the modified DGCNN network. The methodology is tested on the example of the problem of classification of large-scale geospatial objects, such as digital relief models of archaeological sites. A method for accurate registration of objects (FCIP) in the class of affine transformations using geometric and color features was proposed. The results of computer modeling of the proposed methodology based on FICP+DGCNN*+YOLO were presented and discussed. The methodology has theoretical and applied significance not only for the decryption and research of archaeological sites, but also for many applications of digital information processing and robotics in general.
Original languageEnglish
Title of host publicationAnalysis of Images, Social Networks and Texts (AIST 2023)
PublisherSpringer Nature
Pages294-308
Number of pages15
ISBN (Electronic)978-3-031-54534-4
ISBN (Print)978-3-031-54533-7
DOIs
StatePublished - 12 Mar 2024
Externally publishedYes
Event11th International Conference on Analysis of Images, Social Networks and Texts, AIST 2023 - Ереван, Armenia
Duration: 28 Sep 202330 Sep 2023
https://2023.aistconf.org/program/program/

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume14486 LNCS

Conference

Conference11th International Conference on Analysis of Images, Social Networks and Texts, AIST 2023
Country/TerritoryArmenia
CityЕреван
Period28/09/2330/09/23
Internet address

    Research areas

  • 3D semantic segmentation and classification methods, DGCNN, DTM, ICP, object detector

ID: 124116723