Downloads: 4 | Views: 320 | Weekly Hits: ⮙1 | Monthly Hits: ⮙1
Analysis Study Research Paper | Computer Science & Engineering | India | Volume 12 Issue 4, April 2023 | Rating: 4.8 / 10
Deep Learning Model to Revive Indian Manuscripts
Puran Bhat, Kannagi Rajkhowa
Abstract: Sharda is an ancient classic Kal script of Kashmir. It is said to have evolved from the 'Brahmi" also some 1200 years ago. It was much used in undivided Kashmir, North, and Western India (Gilgit), etc, Punjab, Himachal Pradesh, and even up to Central Asia. Devanagari a combination of two words Dev and Nagri( Script of God) is the mother of more than 100 languages spoken in undivided India ( Hind, Kabul, Sind, Pakistan, Barhma, Lanka, and Bhutan, etc ), comprises 13 vowels, 36 consonants, 10 digits. When a vowel is added to a constant the alphabet here gets modified. The Sharda script also supports various compound characters. Due to this recognition is a challenging task. This paper deals with the deep learning module to revive this Indian Manuscript Sharda. Devanagari is an ancient Indian alphabet. This script can be used in Sanskrit, Hindi, Marathi, Bhojpuri, Nepali, Konkani, Sindhi, Marwari, Maithiri language, Pali etc. It is used to write more than 100 language spoken in India and Nepal, including The word "Devanagari" is a combination of two words meaning God and "Nagari" meaning house. Therefore, the word "Sanskrit" means "God's writing". It contains 49 basic letters, 13 sounds and 36 numbers and 10 digits. Also, letters are changed when vowels are added to consonants. The script also supports multiple character combinations. Recognition of the Devanagari script is therefore a challenge and has been an area of research to this day. Researchers have used different types of analysis such as neural networks, support vector machines, particle swarm optimization techniques, and genetic algorithms. This article discusses different classification studies for Devanagari character knowledge.
Keywords: Sharda, Devanagari, Deep Learning
Edition: Volume 12 Issue 4, April 2023
Pages: 1365 - 1368