|
|||||||
|
Machine translation Algorithm for various
Kannada Inscription to Hosagannada Text : Deep Neural Network(DNN) |
|||||||
| Paper Id :
18209 Submission Date :
2023-10-13 Acceptance Date :
2023-10-21 Publication Date :
2023-10-25
This is an open-access research paper/article distributed under the terms of the Creative Commons Attribution 4.0 International, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. DOI:10.5281/zenodo.10184910 For verification of this paper, please visit on
http://www.socialresearchfoundation.com/resonance.php#8
|
|||||||
| |||||||
| Abstract |
In this study, a new methodcalled a deep neural network algorithm (DNN) has been proposed. In this work, symbols of the Hoysala, Ganga, Ashoka, Kadamba, Vijayanagar, Badami and Chalukya periods have been preserved in the structure. Recognizing characters based on optical character recognition (OCR).Recognized characters are translated into valid Kannada script. Test results show greater accuracy and better processing time in analyzing Kannada historical texts. This design uses OCR as a basic software and uses images as a tool to easily write Indian stone inscriptions.In the research article Dr. H S Mohana, “Era identification and recognition of Ganga and Hoysala phase Kannada stone inscriptions using advance recognition algorithm in [2014][1]addressed the methods and technologies. ,In the research article P. Bannigidad and C. Gudada, "Restoration of degraded Kannada handwritten paper inscriptions (Hastaprati) using image enhancement techniques," addressed the image processing techniques in [2017] [2]. “This paper focuses on the development of an method to convert the source sentence[Kannada Inscript] from one natural language of image processing to translate the resultant Hosagganada text with the help of OCR [optical character recognition] using Machine translation”. |
||||||
|---|---|---|---|---|---|---|---|
| Keywords | Artificial Intelligence(AI), Deep-Neural-Network (DNN), Kannada Inscriptions , Optical Character Recognition(OCR), ,Machine Translation Algorithm (ML), Natural Language Processing (NLP),Translation. | ||||||
| Introduction |
The
inscriptions have contributed for Kannada literature and helps to classify it
as Old Kannada,Old Kannada,Old Kannada,intermediate kannada and New
Kannada.Different techniques can be used to create this technique,but we need
strong ideas to create better systems than the existing ones.Kannada was
established during the Mauryan period, which dates back to the 3rd century
BC.It is generally considered to be prehistoric Kannada.over the rule of
various dynasty from Ganga to Rashta Kutas toChalukyas,"Halegannada"
slowly transformed into "Nadugannada" and saw changes in the
scenario.The"Nadugannada"of the Hoysalas,Vijayanagar and Mysore
dynasties underwent many changes over a period of five centuries and is now classified
as"Hosagannada".In the following research paper Rajithkumar B
K1,Dr.H.S.Mohana,used “Template Matching Method for Recognition of Stone
Inscripted Kannada Characters of Different Time Frames based on Correlation
Analysis” on June [2014] [1]Effective
communication training achieves the goal of producing better translations that
provi-de better accuracy,for Kannada inscription to hosagannda text is a ground
breaking initiative that aims to preserve and unlock the linguistic and
cultural treasures embedded within Kannada inscript literature. Kannada
inscript, an ancient writing system, poses significant challenges for
comprehension and accessibility due to variations, complex character mappings,
and linguistic context. This algorithm harnesses the power of artificial
intelligence and machine learning to accurately translate kannada inscript into
modern Kannada, enabling researchers, historians, and language enthusiasts to
decipher and explore the wealth of knowledge and cultural heritage contained
within these inscript texts. By bridging the gap between the past and the
present, this algorithm offers a transformative solution for understanding and
preserving the rich legacy of Kannada inscript. we hope the datasets launched as part of this undertaking will make contributions definitely toward forwarding studies inside the system translation of underneath-resourced languages.
1.1
Terminologies and notations Based
on the various research paper the following article “A Survey of Deep Learning
Techniques for Neural Machine Translation” by
S Yang [ 2020][1] addressed the methods and technologies of DNN(Deep
neural network) Natural
language processing (NLP) combines computational modeling (modeling of human
language) with mathematics, machine learning,and deep learning models. 1.
Natural language processing is the driving force behind machine intelligence in
many applicati--ons today. 2.
Natural language learning is a language with many subfields, such as artificial
intelligence (AI), 3. Machine learning (ML), and Deep learning (DL) such as deep neural networks (DNN). 4. Deep learning (DL) is a subset of ML that consists of algorithms trained to mimic the human brain and perform tasks such as pattern recognition.
Fig.3 Deep Neural Network(DNN) subfield of Deep-Learning(DL) |
||||||
| Objective of study |
This paper offers a top level view on machine translation of Dravidian languages .The following research paper “Machine translation Algorithm for various Kannada Inscription To Hosagannada Text : Using Deep Neural Network (DNN)” used totranslate the text for of kannada inscriptions belongs to historical Kannada Characters. |
||||||
| Review of Literature | 2.1.
Advent to the review This
method addresses the role of knowledge division
in cognitive behavior of research. Optical
Character Recognition (OCR) software from Google [2015]
https://opensource.com/life/15/9/open-sourceextract-text-images: is an example
of an OCR workflow. Use
Kannada notation to show complex results after analysis. 2.2.
Inscription analyzing guide techniques
and Use of OCR
The OCR is used to recognize and translate ancient various Kannada inscription languages. The OCR module contains the technologies of Machine Learning algorithm and Deep-Neural Network(DNN),high accuracy optical character recognition has been report in the prose for nonIndic language such as English, which have very small characters and fewer structural problems.By research review in the paper article Samuel Lukas, Ridvi avyodri “Optical Character Recognition (OCR) for Text Recognition and its Post-Processing Method: A Literature Review” has approached the methods on 14 November [2022][1].Although there are currently several conventional OCR methods for Kannada text,R. Mahesh K. Sinha said in a research article titled "Journey from Hindi Word Processing to Indian Language Processing"in The One Month [2009] [2] that there are almost no studies to do.It was published many years ago in the field of analysis and interpretation of ancient texts,which is a more difficult field than the translation of texts into their current form. |
||||||
| Methodology | In
the research paper Amruta B.Patil,J.A.shaikh,has used OTSU Thresholding Method
for Flower Image Segmentation, on [May 2016] Available
at:http://www.ijceronline.com/papers/Vol6_issue5/A06 50106.pdf DNN
Kannada Inscription To Text Input: Old Kannada Inscript Image Output: Translating into Kannada text |
||||||
| Analysis | Hale-gannadaa
Lipi Data storage(Ashoka-Era)1)
3)
Fig
4.
Architecture of a Deep Neural Network (DNN) Machine-Translation 3.1 Facts series with the aid of Preprocessing the input
image. 1.1 Test engine KOW values for compounds and simple
molecular pipeline input system strings are written to be suitable for building
QSPR models. The string here is sufficient to represent the information
of the molecular structure. 3.2 Function extraction via person areas from the
preprocessed photograph. 2.1 The concatenated string is used to create a list of number
vectors according to the proposed atomic insertion algorithm used by atomic
signature. Vectors define molecular structures and represent their properties. “Comparing the OCR Accuracy Levels of Bitonal and Greyscale
Images” [online] available at http://www.dlib.org/dli
b/march09/powell/03powell.html techniques are referenced and applied for the
following kannada inscripts Translation from halegannada To hosagannada text. 3.4
Information processing. 3.1 These sequences were transformed into classical molecular
signatures using classical molecular graph theory [58]. Based on this, these
features are tuned into a tree-LSTM network whose purpose is to generate vector as input. 3.5 Version training. 4.1.After receiving the input from the TreeLSTM network,
BPNN supports the matching process and works backwards to learn the QSPR model
of interest. During the training process, the parameters are adjusted to take
advantage of the disadvantages of the DNN model,and finally the performance of
the SQSPR model is recorded for KOW prediction. 3.5 Performance evaluation. 5.1.According to the QSPR model, general ability is
evaluated by estimating the quality of external data. And evaluate the exterior
competition of the QSPR form by comparing it with the official estimate
model. 3.6 Output the converted Kannada Text.
quit set of rules |
||||||
| Result and Discussion |
4.1.1) The Vikarama Era
4.1.2) The Saka Era
4.1.3) The Gupta- Valabhi Era
The development of machine translation is a way of converting
sentences from one language to another with the help of computer systems,Our
machine translation system for old kannada Inscript to modern Kannada offers
highly accurate translations between halegannada and hosagannada, very quickly,
and at no cost to users.All the above mentioned kannada old inscripts provide
image to text translation ,our
some of developed dataset from image
processing to Text Translation System
that has addressed various old inscripts
(Halegannada) to hosagannada Text translation through OCR(optical character
recognization),and Deep Neural Network(DNN) method Of Machine Translation
algorithm.
Output is as shown below:
Fig.6
Detection of Kannada Shasana for obtaining Output
Output: Halmidi Shasana |
||||||
| Conclusion |
Inside the improvement of an gadget Translation set of rules for converting antique Kannada inscript to Kannada text represents a considerable development in retaining and decoding the linguistic and culture legacy of kannada inscript, It is used to support further research within this framework.This algorithm, fueled by the power of artificial intelligence and machine learning, accurately converts archaic Kannada inscript into modern Kannada text,opening avenues for researchers,historians and language enthusiasts to access and comprehend ancient Kannada literature.By addressing challenges related to variations,context-dependent mappings,and efficient conversion,this algorithm bridges the gap between the past and the present,unlocking the rich heritage embedded within Kannada inscript and fostering a deeper understanding of Kannda language and its historical roots. |
||||||
| References | 1.
Dr.H.S.Mohana, “Era identification and recognition of Ganga and Hoysala phase
Kannada stone inscriptions using advance recognition algorithm”, 2014
International Conference on Control, Instrumentation, Communication and
Computational Technologies(ICCICCT) 2.
P. Bannigidad and C. Gudada, "Restoration of degraded Kannada handwritten
paper inscriptions (Hastaprati) using image enhancement techniques," 2017
International Conference on Computer Communication and Informatics (ICCCI),
Coimbatore, pp. 1-6, (2017). 3.
Rajithkumar B K1, Dr. H.S. Mohana, “Template Matching Method for Recognition of
Stone Inscripted Kannada Characters of Different Time Frames based on
Correlation Analysis”, International Journal of Computer and Electronics
Research Volume 3, Issue 3, June 2014 4.
“A Survey of Deep Learning Techniques for Neural Machine Translation”by SYang [
2020][1] addressed the methods and technologies of DNN(Deep neural network) 5.
“Google's Optical Character Recognition (OCR) software “[2015]
https://opensource.com/life/15/9/open-sourceextract-text-images 6.
Amruta B. Patil, J.A.shaikh, “OTSU Thresholding Method for Flower Image
Segmentation”, [May 2016] [online].Available at :
http://www.ijceronline.com/papers/Vol6_issue5/A06 50106.pdf 7.“Comparing
the OCR Accuracy Levels of Bitonal and Greyscale Images”
[online]Available:http://www.dlib.org/dli b/march09/powell/03powell.html. 8.Baoguang
Shi. “An end-to-end trainable neural network for image-based sequence
recognition and its application to
scene text recognition”. CoRR. 2015. 9.
Youssouf C. Feature design and lexicon reduction for efficient off-line
handwriting recognition Ph.D.thesis 2016. 10.
Mohana,H.S.,and B.K.Rajithkumar."Era identification and recognition of
Ganga and Hoysala phase Kannada stone inscriptions characters using advance
recognition algorithm." In 2014 International Conference on Control,
Instrumentation, Communication and Computational Technologies (ICCICCT). IEEE,
2014. 11.
M.Mahalakshmi, MalathiSharavanan, “Ancient tamil script and recognition and
Translation using labview”, International conference on Communication and
Signal Processing, April 3-5, 2013, India 12.“A
Journey from Indian Scripts Processing to Indian Language Processing” R. Mahesh
K. Sinha Indian Institute of Technology, Kanpur annals--sinha-jan 2009 13.
“A Comprehensive Guide to Convolutional Neural Networks- the EL15 way” [online]
Available: https://towardsd
atascience.com/a-comprehensive-guide-to-convolutional-neuralnetworks-the-eli5-way-3bd2b1164a53. 14.
“Keras: The Python Deep Learning library”, [online] Available:
https://keras.io/. 15.
“An end-to-end open source machine learning plat- form”, [online] Available:
https://www.tensorflow.org/ 16. L. Torrey and J. Shavlik, “Transfer
Learning”, Appears in the Handbook of
Research on Machine Learning Applications, published by IGI Global,
edited by E. Soria, J. Martin, R. Magdalena, M. Martinez and A. Serrano, 2009. 17.
T. Manigandan, V. Vidhya, V. Dhanalakshmi and B. Nirmala, "Tamil character
recognition from ancient epigraphical inscription using OCR and NLP," 2017
International Conference on Energy, Communication, Data Analytics and Soft
Computing (ICECDS), Chennai, 2017, pp. 1008-1011. 18.OpenCv
, OpenCv python tutorials , [2014] [online].Available at :
http://docs.opencv.org/3.0- beta/doc/py_tutorials/py_tutorials.html 19.
“Grzegorzgwardys , Convolutional Neural Networks backpropagation: from
intuition to derivation”, April 22, [2016],[online].Available at :
https://grzegorzgwardys.wordpress.com/2016/04/22 /8/ 20. Pooja Takur, Prashant Jordon “Django: Developing web using Python” 24 July 2023 IEEE Xplore 23485055 [online] available at: https://ieeexplore.ieee.org/abstract/document/10183246 21. Samuel Lukas, Ridvi avyodri “Optical Character Recognition (OCR) for Text Recognition and its Post-Processing Method: A Literature Review” 14 November 2022 IEEE Xplore 22295656 [online] available at: https://ieeexplore.ieee.org/document/9935961 |
||||||
2)
