Τροποποίηση ή προσθήκη δεδομένων σε αυτήν τη σελίδα

Алгоритмічна Реалізація Цифрових Методів Розпізнавання Нумізматичних Об’єктів

Μεταφρασμένος τίτλος Algorithmic Implementation of the Digital Methods of the Recognition of the Numismatic Objects
Συγγραφείς Natalia Pasichnyk, Renat Rizhniak
Σημοσιευμένο σε Український нумізматичний щорічник, Volume 8 (2024)
The Ukrainian Numismatic Annual
Σελίδες 340-357 (18 σελίδες)
Λήψη https://doi.org/10.31470/2616-6275-2024-8-340-357
Αριθμός
N#
L143173
 
coin, recognition, theory, image, method, algorithmic implementatio, mathematical formulation, software implementation

Περίληψη

The article analyzes the practical implementation of the general problem of using digital technologies in the humanities and cultural sciences to improve accessibility, understanding and interaction. The historical field of research became a convenient field of activity, first for humanities informatics, and later for digital humanities. Nowadays, numismatics as a field of historical research actively uses the digital models of content analysis, mathematical, statistical and probabilistic models, artificial intelligence theory, models of mathematical analysis, analytical geometry and topology. In addition, the tasks of digital recognition of the numismatic material are gaining popularity in numismatics, making the methods of recognizing the images as a relevant subject of practical application for the development of numismatic research. The purpose of the article is to study and systematize the peculiarities of algorithmic implementation of the digital methods of numismatic material recognition. The authors conducted a literature review on the use of digital technologies in numismatics, which covers several key aspects: database creation and management; use of Internet resources for research and collecting; market analysis and coin value assessments; automated coin recognition. The article analyzes the most popular existing applications for coin recognition, classifies the methods of pattern recognition theory (which is the theoretical basis of all intelligent digital systems endowed with the functionality of identifying numismatic objects), and studies their algorithmic implementation. This allowed us to formulate the following conclusions. Firstly, the analysis of the most popular implementations of digital recognition of numismatic objects, which are oriented to use both on stationary devices and mobile devices, showed that the features of their main functionality are reduced to solving the actual problems of coin recognition, and the availability of additional functionality provides users with the opportunity to rationally choose between different products. Secondly, the following groups of generalized methods are used in the practical implementation of the theory of pattern recognition, that is the theoretical basis of all intelligent digital systems endowed with the functionality of identifying numismatic objects: the classical methods, machine learning metho learning methods. Each of the groups of the named generalized methods contains a set of specific methods that determine a certain format and algorithm for working with data. Thirdly, we define the mathematical formulation of the problem of a pattern recognition in a general version, and also reveal the algorithmic and mathematical structures of the implementation of the most popular methods of numismatic object recognition – the pattern method, the nearest neighbor method, the clustering method, and the convolutional neural network method. For each of the named methods, recommendations were provided regarding their software implementations in relevant programming languages and libraries. A follow-up study may involve the determination of algorithmic implementations of the other methods of pattern recognition theory for the identification of numismatic material and the study of options for writing codes to implement the described methods and analyze their performance for various equipment.ds, deep learning methods, and unsupervised

Συμβολή στον κατάλογο

Τροποποίηση σελίδας

Προβλήθηκε 3 φορές

Για να αναφέρεις αυτήν τη σελίδα: https://numista.com/L143173 (αντιγραφή μόνιμου συνδέσμου) Ο μόνιμος σύνδεσμος αντιγράφηκε.