WITHDRAWN: A deep neural network based shape unification to define a 3-Dimensional shape

Autor:	Sushma Jaiswal, K.K. Arun, A. Harshavardhan, V.R. Navaneeth
Rok vydání:	2021
Předmět:	010302 applied physics Theoretical computer science Artificial neural network Unification Computer science 0103 physical sciences 02 engineering and technology 021001 nanoscience & nanotechnology 0210 nano-technology Object (computer science) 01 natural sciences Abstraction (linguistics)
Zdroj:	Materials Today: Proceedings.
ISSN:	2214-7853
DOI:	10.1016/j.matpr.2020.12.1130
Popis:	The ability to define complex in a clear but representative manner Shape is a useful method in machine learning, known as shape abstraction. A deep neural network without supervision that forecasts a number Primitive forms for a three-dimensional arbitrary input. The shape of the aircraft and chair models is built and educated. Any changes to correct unfair outcomes. This is the network can predict primitive sets similar to the form of Input. The primitive sets are not parsimonious, however, i.e. The same part of a can be defined with several primitives Item object. A solution to this problem is suggested by adding a punitive purpose for primitive overlapping, but not generating the effect you like. Effects of direct application of both the network and the improvements are implemented and discussed.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::d7d31cfb5a81bff274f42b42d87828ba https://doi.org/10.1016/j.matpr.2020.12.1130 Zobrazit plný text záznamu Full Text from ScienceDirect