Does Deeper Network Lead to Better Accuracy: A Case Study on Handwritten Devanagari Characters

Chakraborty, Bappaditya; Shaw, Bikash; Aich, Jayanta; Bhattacharya, Ujjwal; Parui, Susanta Kumar

doi:10.1109/das.2018.72

Cited by 29 publications

(8 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A specific group of cells is combined into blocks in which normalization is performed. After normalization, values in the block are combined to form a single feature vector [10]. The features extracted by the extractor are applied to any classifier like ANN and SVM for the classification.…”

Section: Artificial Neural Network (Ann)mentioning

confidence: 99%

“…The features extracted by the extractor are applied to any classifier like ANN and SVM for the classification. The paper [10] has implemented HOG as a feature extractor along with ANN as a classifier. It achieved an accuracy of 82.66% using the ISI handwritten character database with input image of size 32 × 32.…”

Section: Artificial Neural Network (Ann)mentioning

confidence: 99%

“…BLSTM is a classifier so it needs a feature extractor to provide feature vector as input to it. BLSTM can be used with CNN as well as HOGbased feature descriptor which is proposed in the paper [10]. It has achieved accuracy of 94.56% and 79.54 with CNN and HOG as a feature extractor, respectively.…”

Section: Bidirectional Long Short-term Memory (Blstm)-based Recognitionmentioning

confidence: 99%

See 2 more Smart Citations

Implementation of Residual Network (ResNet) for Devanagari Handwritten Character Recognition

Mhapsekar

Mhatre

et al. 2020

Algorithms for Intelligent Systems

View full text Add to dashboard Cite

The study in the field of optical character recognition can be traced back to mid-1940s and has ever since gaining the attention of various industries and sectors. Optical character recognition has been highly used in banks, post offices, libraries, and publishing houses. The main challenge in OCR is handwritten character recognition. The research on handwritten character recognition began in the late 1960s, and at that time, only the handwritten numeric characters were addressed by the system [1]. Over the years, the technological approach for solving this problem has developed, thus improving the accuracy of the system [2]. Handwritten OCR for the English language which comes under the Latin script has almost developed into a full-fledged system. The research on handwritten OCR for Devanagari script is very limited compared to Latin script. Devanagari script includes languages like Hindi, Marathi, and Nepali. The most widely used Devanagari script language is Hindi with over 500 million people using this language [3].In Devanagari script, there are 13 vowels and 36 consonants as shown in Fig. 1. There are 14 modifiers in Devanagari, out of which 11 are of vowels and 3 are of

show abstract

Section: Artificial Neural Network (Ann)mentioning

confidence: 99%

Section: Artificial Neural Network (Ann)mentioning

confidence: 99%

Section: Bidirectional Long Short-term Memory (Blstm)-based Recognitionmentioning

confidence: 99%

See 1 more Smart Citation

Implementation of Residual Network (ResNet) for Devanagari Handwritten Character Recognition

Mhapsekar

Mhatre

et al. 2020

Algorithms for Intelligent Systems

View full text Add to dashboard Cite

show abstract

“…Bappaditya Chakraborty et al, [8] presented a case study on recognition of handwritten Devanagari Characters. In this study, they have implemented deep neural networks using CNN and Bidirectional Long Short-Term Memory (BLSTM) layers in between the convolutional and the fully connected part of CNN networks and obtained accuracy of 96.09%.…”

Section: Related Workmentioning

confidence: 99%

Recognition of Online Handwritten Isolated Kannada Characters using PCA and DTW

Prasad G*,

Hegde

2019

IJRTE

View full text Add to dashboard Cite

Handwriting is a natural means of documentation and communication for several years. Human beings communicating with computers through handwritten input would be the best and easiest way of exchanging the information. It is difficult to input data for computers for Indian language scripts because of their complex typing nature. This paper focuses on exploring performance of Principal Component Analysis (PCA) and Dynamic Time Wrapping (DTW) approaches for recognizing online handwritten isolated Kannada characters. Methodology proposed in this paper is writer independent model which recognizes basic 50 Kannada characters including 16 vowels and 34 consonants.

show abstract

“…Authors have put forward a study regarding the effect of depth of Convolutional Neural Network (CNN) architecture on recognition accuracy of handwritten Devanagari characters. It has been concluded that CNN-BLSTM hybrid architecture has dominated over state-of-the-art Devanagari character recognition (Chakraborty et al, 2018). Mathur et al(2019) have proposed an integrated system of OCR and text-to-speech conversion to help visually impaired people to read a document.…”

Section: Introductionmentioning

confidence: 99%

Application of Zonal and Curvature Features to Numerals Recognition

Prasad

2021

ijsrtm

View full text Add to dashboard Cite

Purpose of the study: The purpose of this work is to present an offline Optical Character Recognition system to recognise handwritten English numerals to help automation of document reading. It helps to avoid tedious and time-consuming manual typing to key in important information in a computer system to preserve it for a longer time. Methodology: This work applies Curvature Features of English numeral images by encoding them in terms of distance and slope. The finer local details of images have been extracted by using Zonal features. The feature vectors obtained from the combination of these features have been fed to the KNN classifier. The whole work has been executed using the MatLab Image Processing toolbox. Main Findings: The system produces an average recognition rate of 96.67% with K=1 whereas, with K=3, the rate increased to 97% with corresponding errors of 3.33% and 3% respectively. Out of all the ten numerals, some numerals like ‘3’ and ‘8’ have shown respectively lower recognition rates. It is because of the similarity between their structures. Applications of this study: The proposed work is related to the recognition of English numerals. The model can be used widely for recognition of any pattern like signature verification, face recognition, character or word recognition in another language under Natural Language Processing, etc. Novelty/Originality of this study: The novelty of the work lies in the process of feature extraction. Curves present in the structure of a numeral sample have been encoded based on distance and slope thereby presenting Distance features and Slope features. Vertical Delta Distance Coding (VDDC) and Horizontal Delta Distance Coding (HDDC) encode a curve from vertical and horizontal directions to reveal concavity and convexity from different angles.

show abstract

Does Deeper Network Lead to Better Accuracy: A Case Study on Handwritten Devanagari Characters

Cited by 29 publications

References 20 publications

Implementation of Residual Network (ResNet) for Devanagari Handwritten Character Recognition

Implementation of Residual Network (ResNet) for Devanagari Handwritten Character Recognition

Recognition of Online Handwritten Isolated Kannada Characters using PCA and DTW

Application of Zonal and Curvature Features to Numerals Recognition

Contact Info

Product

Resources

About