Handwritten Numeral Recognition Research Articles

Advancing Multilingual Handwritten Numeral Recognition With Attention-Driven Transfer Learning

Advancing Multilingual Handwritten Numeral Recognition With Attention-Driven Transfer Learning

Design and Development of Handwritten Digit Recognition System Using Machine Learning

In any system ,the wide range of writing styles, handwritten digit recognition has proven to be a challenging task.The wide variety of writing styles has made it challenging to read handwritten numerals.. Numerous research have demonstrated the superior performance of neural networks in the classification of data. The major goal of this work is to compare various existing classification models in order to give effective and trustworthy methods for handwritten numeral recognition. As training sets, a total of 45,000 images with a pixel size of 2828 were employed. The original image was compared to the training sets and images. We also found that the classifier's accuracy increases with the amount of training data. The study's conclusion shows that K-NN maintains an accuracy of 96.7% as opposed to 96.8% while processing data at a rate nearly ten times faster than a neural network.

AttentionMNIST: a mouse-click attention tracking dataset for handwritten numeral and alphabet recognition

Multiple attention-based models that recognize objects via a sequence of glimpses have reported results on handwritten numeral recognition. However, no attention-tracking data for handwritten numeral or alphabet recognition is available. Availability of such data would allow attention-based models to be evaluated in comparison to human performance. We collect mouse-click attention tracking data from 382 participants trying to recognize handwritten numerals and alphabets (upper and lowercase) from images via sequential sampling. Images from benchmark datasets are presented as stimuli. The collected dataset, called AttentionMNIST, consists of a sequence of sample (mouse click) locations, predicted class label(s) at each sampling, and the duration of each sampling. On average, our participants observe only 12.8% of an image for recognition. We propose a baseline model to predict the location and the class(es) a participant will select at the next sampling. When exposed to the same stimuli and experimental conditions as our participants, a highly-cited attention-based reinforcement model falls short of human efficiency.

Handwritten Numeral Recognition Integrating Start–End Points Measure with Convolutional Neural Network

Convolutional neural network (CNN) based methods have succeeded for handwritten numeral recognition (HNR) applications. However, CNN seems to misclassify similarly shaped numerals (i.e., the silhouette of the numerals that look the same). This paper presents an enhanced HNR system to improve the classification accuracy of the similarly shaped handwritten numerals incorporating the terminals points with CNN’s recognition, which can be utilized in various emerging applications related to language translation. In handwritten numerals, the terminal points (i.e., the start and end positions) are considered additional properties to discriminate between similarly shaped numerals. Start–End Writing Measure (SEWM) and its integration with CNN is the main contribution of this research. Traditionally, the classification outcome of a CNN-based system is considered according to the highest probability exposed for a particular numeral category. In the proposed system, along with such classification, its probability value (i.e., CNN’s confidence level) is also used as a regulating element. Parallel to CNN’s classification operation, SEWM measures the start-end points of the numeral image, suggesting the numeral category for which measured start-end points are found close to reference start-end points of the numeral class. Finally, the output label or system’s classification of the given numeral image is provided by comparing the confidence level with a predefined threshold value. SEWM-CNN is a suitable HNR method for Bengali and Devanagari numerals compared with other existing methods.

BANGLA HANDWRITTEN NUMERAL RECOGNITION USING DEEP CONVOLUTIONAL NEURAL NETWORK

The recognition of Bangla handwritten numerals (BHNR) has recently emerged as a very interesting area for machine learning and pattern recognition research. Recently, the technology for character or object recognition has also advanced. Bangla handwritten number recognition can serve as a foundation for creating an Optical Character Recognition (OCR) in the Bangla language. However, the lack of a sizable and accurate dataset makes Bangla's handwritten numeral recognition study insufficient in contrast to that of other well-known languages. Similar to MNIST for English digits, NumtaDB is by far the largest dataset collection for handwritten digits in the Bangla language. The most used datasets for the recognition of Bangla handwritten numerals in the past were NumtaDB, CMATERdb, and ISI. The majority of approaches now in use rely on feature extraction and a few outdated machine learning algorithms. Although some approaches operate quickly enough to meet practical demands, they are not always accurate. Even while certain techniques work quite well for languages other than Bangla, they still require improvement. Convolutional Neural Networks (CNN), in particular, are demonstrating excellent achievements in this discipline with higher accuracy. In this work, we’ve used custom CNN architectures to build our model to recognize digits using all the existing datasets with a high degree of accuracy. Our CNN model shows an average of 98% accuracy recognizing Bangla numeric in respect of above datasets. We have cross verified our model with mixed datasets and the result is also promising.

Digital Recognition Methods Based on Deep Learning

In this paper, the K-nearest neighbor algorithm and the convolutional neural network will be used to train the handwritten digit recognition model, respectively. To establish a reasonable model structure, and through the training data, the model can learn to reflect ten different handwritten number features and finally give the probability of predicting number corresponding to the likelihood of each number. Taking the learning process of the handwritten numeral recognition algorithm based on deep learning as a clue, from deep learning to convolutional neural network, from simple to deep, the relevant basic concepts, model construction, and training process of deep learning are learned and understood. Finally, the deep learning framework uses MNIST as the training dataset to train a model with high recognition rate and then combines it with Open CV technology to realize the identification of handwritten numbers. A reasonable model structure is used to accurately identify the handwritten numbers in the test set. The neural network of deep learning is established with TensorFlow to realize the classification and recognition of handwritten numbers. Various deep learning methods such as CNN and KNN are learned and compared to complete the construction of deep learning architecture. The MNIST dataset was preprocessed, features extracted, and identified. The program is to complete the training of neural network and the recognition of numbers in the image, the recognition results of deep learning methods used are counted and analyzed, and the recognition rates of two different methods are compared to find ways to optimize these methods and improve the recognition rate.

Handwritten numeral recognition through superimposition onto printed form

Handwriting is, eventually, a variation of the printed forms where the characters are little larger, smaller, angled and deformed than the printed forms. The small changes in handwriting define the parameters of the character to be recognized. Handwritten numeral recognition (HNR) poses significant challenges due to the deformations and other variations. This study proposes a new notion of HNR on the hypothesis that the handwritten numerals are distinct deformations of the printed forms, which leads to easier recognition task with higher accuracy when superimposing handwritten numeral images onto the corresponding printed numeral images. In the proposed HNR, auto-encoder and convolutional auto-encoder have been adapted for the superimposing task that transform HNIs into PNIs, while neural network and convolutional neural network are employed for classification of PNIs. The superimposition method reduces the computational overhead. Moreover, this method employs simple pre-processing withoutfeature extraction whereas the traditional methods employ pre-processing, feature extraction, and recognition with machine learning tools, which add to the computational overhead. The performance of HNRSP has been evaluated for recognizing handwritten numerals of Bengali, Devanagari, and English on benchmark datasets andthe proposed system achieves 99.68%, 99.73%, and 99.62% recognition accuracy for Bengali, Devanagari, and English handwritten numerals, respectively.

Enhanced performance of a reservoir computing system based on a dual-loop optoelectronic oscillator.

Time-delayed reservoir computing (RC) is a brain inspired paradigm for processing temporal information, with simplification in the network's architecture using virtual nodes embedded in a temporal delay line. In this work, a novel, to the best of our knowledge, RC system based on a dual-loop optoelectronic oscillator is proposed to enhance the prediction and classification. The hardware is compact and easy to implement, and only a section of fiber compared to the traditional optoelectronic oscillator reservoir is added to conform the dual-loop scheme. Compared with the traditional reservoir, a remarkable performance of the proposed RC system is demonstrated by simulation on three well-known tasks, namely the nonlinear auto regressive moving average (NARMA10) task, signal waveform recognized task, and handwritten numeral recognition. The parameter optimization in the NARMA10 task is presented with influenced factors. The novel RC system finally obtains a normalized mean square error at 0.0493±0.007 in NARMA10 task, 6.172×10-6 in signal waveform recognized task, and a word error rate at 9% in handwritten numeral recognition with suitable parameters.

Real time handwritten digital image recognition system based on edge computing equipment

With the development of edge computing technology, real-time handwritten numeral recognition system deployed in edge computing devices has a bright future. However, the edge computing equipment has weak computing power and limited storage space, so the mainstream image recognition neural network can not guarantee the real-time performance on low-performance edge devices. To solve this problem, this paper designs a handwritten digit recognition system which is suitable for low performance edge computing devices. The system extracts the effective information such as the position of the number in the input image through the image preprocessing module, and infers through a lightweight neural network specially designed for edge devices. In this paper, the proposed handwritten numeral image recognition system is deployed on the edge computing device Jetson Nano. The experimental data show that the inference speed of our model is 10 times faster than that of the original Tensorflow inference, and 60 times higher than the neural network Mobilenet specially designed for mobile devices. At the same time, with the increase of input video resolution, the FPS of the system does not decline significantly, which can meet the needs of most edge tasks. Finally, the system also provides design and deployment experience for other edge AI tasks.

Improved Model Configuration Strategies for Kannada Handwritten Numeral Recognition

Handwritten numeral recognition has been an important area in the domain of pattern classification. The task becomes even more daunting when working with non-Roman numerals. While convolutional neural networks are the preferred choice for modeling the image data, the conception of techniques to obtain faster convergence and accurate results still poses an enigma to the researchers. In this paper, we present new methods for the initialization and the optimization of the traditional convolutional neural network architecture to obtain better results for Kannada numeral images. Specifically, we propose two different methods- an encoderdecoder setup for unsupervised training and weight initialization, and a particle swarm optimization strategy for choosing the ideal architecture configuration of the CNN. Unsupervised initial training of the architecture helps for a faster convergence owing to more task-suited weights as compared to random initialization while the optimization strategy is helpful to reduce the time required for the manual iterative approach of architecture selection. The proposed setup is trained on varying handwritten Kannada numerals. The proposed approaches are evaluated on two different datasets: a standard Dig-MNIST dataset and a custom-built dataset. Significant improvements across multiple performance metrics are observed in our proposed system over the traditional CNN training setup. The improvement in results makes a strong case for relying on such methods for faster and more accurate training and inference of digit classification, especially when working in the absence of transfer learning.

Multilingual handwritten numeral recognition using a robust deep network joint with transfer learning

Numeral recognition plays a crucial role in creating automated systems such as posting address sorting and license plate recognition. Nowadays, numeral recognition systems which have the capability of recognising multiple languages are highly beneficial due to growing international correspondence and transactions, especially in multilingual countries where several languages are used simultaneously. Therefore, handwritten numeral recognition is more challenging than printed numeral recognition due to having different and complex handwriting styles. Hence, developing a multilingual handwritten system is considered as an important and debatable issue. We address this issue by proposing a language-independent model based on a robust CNN. Our proposed model is composed of language recognition and digit recognition, which aims to handle the recognition of multi-script images. We used transfer-learning in the proposed system to enhance the image quality and consequently the recognition performance. Extensive experiments were conducted to verify the effectiveness of both language and digit recognition procedures. The proposed system was tested with six different languages. The results showed an average accuracy of up to 99.8% for recognising various languages and the associated digits. The robustness and design procedure of the proposed model created a cost-effective extension for recognition of handwritten numerals in other languages.

Design of Handwritten Numeral Recognition System Based on BP Neural Network

In the case of pattern recognition, handwritten numeral recognition is an important research topic in pattern recognition, and it has a very wide application in today’s information society. However, the research on numeral recognition is still in the development stage, and the recognition effect is not ideal. A handwritten numeral recognition method based on BP neural network is proposed. Firstly, the image is grayed, binarized, smoothed, denoised and normalized to extract the pixel value; Then the designed BP neural network is trained, compared with the expected results and expected structure, and the BP neural network is adjusted and modified; Finally, the trained neural network is obtained. Experiments show that the accuracy of this method for handwritten digit recognition is 85.88%.

TenzinNet for handwritten Tibetan numeral recognition

TenzinNet for handwritten Tibetan numeral recognition

A new feature extraction approach for script invariant handwritten numeral recognition

AbstractHandwritten numeral recognition is a challenging research problem because of the enormous varieties of styles in which human beings write the numerals. Several researchers have tried to find solutions to this problem with exceptional recognition accuracies. However, most of these solutions have been dedicated to single script numerals. Such methods are inappropriate for multi‐lingual nations such as India where a large number of scripts are used. Keeping this issue in mind, a new feature descriptor named symbolization of binary images (SBI) is introduced here for the recognition of handwritten numerals of different scripts. Effectiveness of SBI is supported with experiments showing its script‐invariant nature. Classification of numerals using a multiclass support vector machine (SVM) classifier yields the recognition accuracies of 98.18, 96.22, 96.52, and 95.53% on datasets of numerals written in four popular scripts of the world: Arabic, Bangla, Devanagari, and Latin, respectively. This scheme has also been extended to the situation when the script used is not known a priori or the numerals written in a document belong to pairs of mixed scripts of {Arabic, Devanagari, Bangla} with Latin producing recognition rates of 92.97, 91.25, and 91.67%, respectively. When all four scripts are mixed, the recognition rate is still 90.98% overall. Encouraging outcomes suggest that the proposed SBI feature descriptor can recognize numerals invariant of the script class.

A Robust Handwritten Numeral Recognition Using Hybrid Orthogonal Polynomials and Moments.

Numeral recognition is considered an essential preliminary step for optical character recognition, document understanding, and others. Although several handwritten numeral recognition algorithms have been proposed so far, achieving adequate recognition accuracy and execution time remain challenging to date. In particular, recognition accuracy depends on the features extraction mechanism. As such, a fast and robust numeral recognition method is essential, which meets the desired accuracy by extracting the features efficiently while maintaining fast implementation time. Furthermore, to date most of the existing studies are focused on evaluating their methods based on clean environments, thus limiting understanding of their potential application in more realistic noise environments. Therefore, finding a feasible and accurate handwritten numeral recognition method that is accurate in the more practical noisy environment is crucial. To this end, this paper proposes a new scheme for handwritten numeral recognition using Hybrid orthogonal polynomials. Gradient and smoothed features are extracted using the hybrid orthogonal polynomial. To reduce the complexity of feature extraction, the embedded image kernel technique has been adopted. In addition, support vector machine is used to classify the extracted features for the different numerals. The proposed scheme is evaluated under three different numeral recognition datasets: Roman, Arabic, and Devanagari. We compare the accuracy of the proposed numeral recognition method with the accuracy achieved by the state-of-the-art recognition methods. In addition, we compare the proposed method with the most updated method of a convolutional neural network. The results show that the proposed method achieves almost the highest recognition accuracy in comparison with the existing recognition methods in all the scenarios considered. Importantly, the results demonstrate that the proposed method is robust against the noise distortion and outperforms the convolutional neural network considerably, which signifies the feasibility and the effectiveness of the proposed approach in comparison to the state-of-the-art recognition methods under both clean noise and more realistic noise environments.

Recognition of Handwritten English Numerals Based on Combining Structural and Statistical Features

Generally, pattern recognition considered a strong challenge in many information processing research fields. The aim of this paper is to propose a highly accurate model for recognizing a handwritten English numeral through efficiently extracting the most valuable features of a certain handwritten numeral or digit. The handwritten English Numerals Recognition Model (HENRM) is proposed in this paper. The features extraction of the proposal based on combining both statistical and structural features of the certain numeral sample image. Mainly, the proposed HENCM has four phases which are image acquisition, image preprocessing, features extraction, and classification. In fact, four feature extraction approaches are utilized in this paper, which are the number of intersection points, the number of open-end points, calculation of density feature, and determining the chain code for each of the English numerals. The latter phase gives a features vector of 26-element size to be fed into the classifier that uses the Multi-class Support Vector Machine (MSVM) for the classification process. The experimental results showed that the proposed HENCM exhibits an average recognition rate equals to 97%. Index Terms—Chain Code, Density feature, MSVM, Recognition.

A CNN based Handwritten Numeral Recognition Model for Four Arithmetic Operations

The pandemic of Covid-19 has caused a shift of paradigm of education, from face-to-face to e-learning. E-learning leads to an escalation in digitalization of handwritten documents because it requires submission of homework and assignments through online. To help teachers in checking digitalized handwritten homework, this paper proposes an automatic checking system based on a convolutional neural network (CNN) for handwritten numeral recognition. The CNN is used to recognize four arithmetic operations in mathematical questions consisting of addition, deduction, multiplication and division. The performance CNN in handwritten numeral recognition have been optimized in terms of activation function and gradient descent algorithm. The proposed CNN is also trained and tested with the MNIST handwritten data set. The experimental results show that the recognition accuracy the improved CNN improves to a certain extent as compared to before optimization.

Research on Tibetan Handwritten Numerals Recognition Based on TextCaps Model with Few Training Sample

Research on Tibetan Handwritten Numerals Recognition Based on TextCaps Model with Few Training Sample

CNN-based multilingual handwritten numeral recognition: A fusion-free approach

Numeral recognition plays a vital role in making automated systems like postal address sorting and license plate recognition. In a multilingual country like India, more often, the multiple languages are mixed while writing. Numeral recognizer systems which can handle more than one language are very much useful to recognize numerals of various scripts. Handwritten numeral recognition is much more complicated than the printed one because of the different writing styles. Few multilingual works have been reported earlier, but these systems are either trained for each language individually or fusion of similar shaped classes of performed. In either case, multiple classes exist for a single digit. In this work, we have developed a script independent numeral recognition system for multilingual handwritten digits which is independent of fusion and has only 10 classes corresponding to every single numeric digit. This work is first of its kind in which we have addressed the problems of multilingual numeral recognition systems. Exhaustive experiments are done with numeral datasets of 8 Indic and non-Indic scripts. We have obtained the accuracy of 96.23% collectively for all the eight scripts. This obtained high accuracy is promising and demonstrates the hypothesis that multilingual handwritten numeral recognition is void with CNN.

Language-invariant novel feature descriptors for handwritten numeral recognition

Numeral recognition is treated as a benchmark research problem as this is a basic module for designing a comprehensive optical character recognition system. In this context, unconstrained handwritten numeral recognition is still considered as an open research problem. Most of the feature descriptors found in the literature for the said problem, work well for numeral images written in a particular language. To encounter this shortcoming, in this paper, we have proposed two shape-based feature descriptors, namely Point-Light Source-based Shadow (PLSS) and Histogram of Oriented Pixel Positions (HOPP). We have evaluated the proposed feature descriptors on 10 (9 offline and 1 online) publicly available standard handwritten numeral image datasets written in eight different languages. Besides, to prove the usefulness of the descriptors in real-life scenario, we have considered numeral string images also. We have also shown how the proposed feature descriptors are invariant toward broken, noisy and rotated numeral images. Experimental outcomes soundly prove that the proposed feature descriptors have the ability to estimate the shape of a numeral image almost accurately irrespective of the language in which it is written. Comparison of the proposed feature descriptors with other shape-based as well as texture-based features shows that PLSS and HOPP produce the results which are analogous to state of the art. The code of the proposed feature descriptors can be found at— https://github.com/ghoshsoulib/Numeral-Recognition .