A Novel Siamese Network for Few/Zero-Shot Handwritten Character Recognition Tasks

Deep metric learning is one of the recommended methods for the challenge of supporting few/zero-shot learning by deep networks.It depends on building a Siamese architecture of two homogeneous Convolutional Neural Networks(CNNs)for learning a distance function that can map input data from the input space to the feature space.Instead of determining the class of each sample,the Siamese architecture deals with the existence of a few training samples by deciding if the samples share the same class identity or not.The traditional structure for the Siamese architecture was built by forming two CNNs from scratch with randomly initialized weights and trained by binary cross-entropy loss.Building two CNNs from scratch is a trial and error and time-consuming phase.In addition,training with binary crossentropy loss sometimes leads to poor margins.In this paper,a novel Siamese network is proposed and applied to few/zero-shot Handwritten Character Recognition(HCR)tasks.The novelties of the proposed network are in.1)Utilizing transfer learning and using the pre-trained AlexNet as a feature extractor in the Siamese architecture.Fine-tuning a pre-trained network is typically faster and easier than building from scratch.2)Training the Siamese architecture with contrastive loss instead of the binary cross-entropy.Contrastive loss helps the network to learn a nonlinear mapping function that enables it to map the extracted features in the vector space with an optimal way.The proposed network is evaluated on the challenging Chars74K datasets by conducting two experiments.One is for testing the proposed network in few-shot learning while the other is for testing it in zero-shot learning.The recognition accuracy of the proposed network reaches to 85.6%and 82%in few-and zero-shot learning respectively.In addition,a comparison between the performance of the proposed Siamese network and the traditional Siamese CNNs is conducted.The comparison results show that the proposed network achieves higher recognition results in less time.The proposed network reduces the training time from days to hours in both experiments.

An Optimized Deep Residual Network with a Depth Concatenated Block for Handwritten Characters Classification

Even though much advancements have been achieved with regards to the recognition of handwritten characters,researchers still face difficulties with the handwritten character recognition problem,especially with the advent of new datasets like the Extended Modified National Institute of Standards and Technology dataset(EMNIST).The EMNIST dataset represents a challenge for both machine-learning and deep-learning techniques due to inter-class similarity and intra-class variability.Inter-class similarity exists because of the similarity between the shapes of certain characters in the dataset.The presence of intra-class variability is mainly due to different shapes written by different writers for the same character.In this research,we have optimized a deep residual network to achieve higher accuracy vs.the published state-of-the-art results.This approach is mainly based on the prebuilt deep residual network model ResNet18,whose architecture has been enhanced by using the optimal number of residual blocks and the optimal size of the receptive field of the first convolutional filter,the replacement of the first max-pooling filter by an average pooling filter,and the addition of a drop-out layer before the fully connected layer.A distinctive modification has been introduced by replacing the final addition layer with a depth concatenation layer,which resulted in a novel deep architecture having higher accuracy vs.the pure residual architecture.Moreover,the dataset images’sizes have been adjusted to optimize their visibility in the network.Finally,by tuning the training hyperparameters and using rotation and shear augmentations,the proposed model outperformed the state-of-the-art models by achieving average accuracies of 95.91%and 90.90%for the Letters and Balanced dataset sections,respectively.Furthermore,the average accuracies were improved to 95.9%and 91.06%for the Letters and Balanced sections,respectively,by using a group of 5 instances of the trained models and averaging the output class probabilities.

Optimised CNN Architectures for Handwritten Arabic Character Recognition

Handwritten character recognition is considered challenging compared with machine-printed characters due to the different human writing styles.Arabic is morphologically rich,and its characters have a high similarity.The Arabic language includes 28 characters.Each character has up to four shapes according to its location in the word(at the beginning,middle,end,and isolated).This paper proposed 12 CNN architectures for recognizing handwritten Arabic characters.The proposed architectures were derived from the popular CNN architectures,such as VGG,ResNet,and Inception,to make them applicable to recognizing character-size images.The experimental results on three well-known datasets showed that the proposed architectures significantly enhanced the recognition rate compared to the baseline models.The experiments showed that data augmentation improved the models’accuracies on all tested datasets.The proposed model outperformed most of the existing approaches.The best achieved results were 93.05%,98.30%,and 96.88%on the HIJJA,AHCD,and AIA9K datasets.

Sailfish Optimizer with Deep Transfer Learning-Enabled Arabic Handwriting Character Recognition

The recognition of the Arabic characters is a crucial task incomputer vision and Natural Language Processing fields. Some major complicationsin recognizing handwritten texts include distortion and patternvariabilities. So, the feature extraction process is a significant task in NLPmodels. If the features are automatically selected, it might result in theunavailability of adequate data for accurately forecasting the character classes.But, many features usually create difficulties due to high dimensionality issues.Against this background, the current study develops a Sailfish Optimizer withDeep Transfer Learning-Enabled Arabic Handwriting Character Recognition(SFODTL-AHCR) model. The projected SFODTL-AHCR model primarilyfocuses on identifying the handwritten Arabic characters in the inputimage. The proposed SFODTL-AHCR model pre-processes the input imageby following the Histogram Equalization approach to attain this objective.The Inception with ResNet-v2 model examines the pre-processed image toproduce the feature vectors. The Deep Wavelet Neural Network (DWNN)model is utilized to recognize the handwritten Arabic characters. At last,the SFO algorithm is utilized for fine-tuning the parameters involved in theDWNNmodel to attain better performance. The performance of the proposedSFODTL-AHCR model was validated using a series of images. Extensivecomparative analyses were conducted. The proposed method achieved a maximum accuracy of 99.73%. The outcomes inferred the supremacy of theproposed SFODTL-AHCR model over other approaches.

Research on Handwritten Chinese Character Recognition Based on BP Neural Network

The application of pattern recognition technology enables us to solve various human-computer interaction problems that were difficult to solve before.Handwritten Chinese character recognition,as a hot research object in image pattern recognition,has many applications in people’s daily life,and more and more scholars are beginning to study off-line handwritten Chinese character recognition.This paper mainly studies the recognition of handwritten Chinese characters by BP(Back Propagation)neural network.Establish a handwritten Chinese character recognition model based on BP neural network,and then verify the accuracy and feasibility of the neural network through GUI(Graphical User Interface)model established by Matlab.This paper mainly includes the following aspects:Firstly,the preprocessing process of handwritten Chinese character recognition in this paper is analyzed.Among them,image preprocessing mainly includes six processes:graying,binarization,smoothing and denoising,character segmentation,histogram equalization and normalization.Secondly,through the comparative selection of feature extraction methods for handwritten Chinese characters,and through the comparative analysis of the results of three different feature extraction methods,the most suitable feature extraction method for this paper is found.Finally,it is the application of BP neural network in handwritten Chinese character recognition.The establishment,training process and parameter selection of BP neural network are described in detail.The simulation software platform chosen in this paper is Matlab,and the sample images are used to train BP neural network to verify the feasibility of Chinese character recognition.Design the GUI interface of human-computer interaction based on Matlab,show the process and results of handwritten Chinese character recognition,and analyze the experimental results.

Pashto Characters Recognition Using Multi-Class Enabled Support Vector Machine

During the last two decades signicant work has been reported in the eld of cursive language’s recognition especially,in the Arabic,the Urdu and the Persian languages.The unavailability of such work in the Pashto language is because of:the absence of a standard database and of signicant research work that ultimately acts as a big barrier for the research community.The slight change in the Pashto characters’shape is an additional challenge for researchers.This paper presents an efcient OCR system for the handwritten Pashto characters based on multi-class enabled support vector machine using manifold feature extraction techniques.These feature extraction techniques include,tools such as zoning feature extractor,discrete cosine transform,discrete wavelet transform,and Gabor lters and histogram of oriented gradients.A hybrid feature map is developed by combining the manifold feature maps.This research work is performed by developing a medium-sized dataset of handwritten Pashto characters that encapsulate 200 handwritten samples for each 44 characters in the Pashto language.Recognition results are generated for the proposed model based on a manifold and hybrid feature map.An overall accuracy rates of 63.30%,65.13%,68.55%,68.28%,67.02%and 83%are generated based on a zoning technique,HoGs,Gabor lter,DCT,DWT and hybrid feature maps respectively.Applicability of the proposed model is also tested by comparing its results with a convolution neural network model.The convolution neural network-based model generated an accuracy rate of 81.02%smaller than the multi-class support vector machine.The highest accuracy rate of 83%for the multi-class SVM model based on a hybrid feature map reects the applicability of the proposed model.

Parallel compact integration in handwritten Chinese character recognition

In this paper, a new parallel compact integration scheme based on multi-layer perceptron (MLP) networks is proposed to solve handwritten Chinese character recognition (HCCR) problems. The idea of metasynthesis is applied to HCCR, and compact MLP network classifier is defined. Human intelligence and computer capabilities are combined together effectively through a procedure of two-step supervised learning. Compared with previous integration schemes, this scheme is characterized with parallel compact structure and better performance. It provides a promising way for applying MLP to large vocabulary classification.

A New Linguistic Decoding Method for Online Handwritten Chinese Character Recognition

This paper presents a new linguistic decoding method for online handwritten Chinese character recognition. The method employs a hybrid language model which combines N-gram and linguistic rules by rule quantification technique. The linguistic decoding algorithm consists of three stages: word lattice construction, the optimal sentence hypothesis search and self-adaptive learning mechanism. The technique has been applied to palmtop computer's online handwritten Chinese character recognition. Samples containing millions of characters were used to test the linguistic decoder. In the open experiment, accuracy rate up to 92% is achieved, and the error rate is reduced by 68%.

Model based odia numeral recognition using fuzzy aggregated features

In this paper, an efficient scheme for recognition of handwritten Odia numerals using hidden markov model （HMM） has been proposed. Three different feature vectors for each of the numeral is generated through a polygonal approximation of object contour. Subsequently, aggregated feature vector for each numeral is derived from these three primary feature vectors using a fuzzy inference system. The final feature vector is divided into three levels and interpreted as three different states for HMM. Ten different three-state ergodic hidden markov models （HMMs） are thus constructed corresponding to ten numeral classes and parameters are calculated from these models. For the recognition of a probe numeral, its log-likelihood against these models are computed to decide its class label. The proposed scheme is implemented on a dataset of 2500 handwritten samples and a recognition accuracy of 96.3% has been achieved. The scheme is compared with other competent schemes.