Recently, a few works on recognition of handwritten Bangla characters have been reported in the literature. However, there is scope for further research in this area. In the present article, results of our recent study on recognition of handwritten Bangla basic characters will be reported. This is a 50 class problem since the alphabet of Bangla has 50 basic characters. In this study, features are obtained by computing local chain code histograms of input character shape. Comparative recognition results are obtained between computation of the above feature based on the contour and one-pixel skeletal representations of the input character image. Also, the classification results are obtained after down sampling the histogram feature by applying Gaussian filter in both these cases. Multilayer perceptrons (MLP) trained by back propagation (BP) algorithm are used as classifiers in the present study. Near exhaustive studies are done for selection of its hidden layer size. An analysis of the misclassified samples shows an interesting error pattern and this has been used for further improvement in the recognition results. Final recognition accuracies on the training and the test sets are respectively 94.65% and 92.14%.

On Recognition of handwritten bangla characters

Uncertainty quantification in skin cancer classification using three-way decision-based Bayesian deep learning

https://eartharxiv.org/repository/object/134/download/274/

Fully automated carbonate petrography using deep convolutional neural networks

In this paper, we tackle the twins identification problem; a challenging task in biometric authentication. The experiments are carried out using deep learning techniques and for that we have proposed a Convolutional Siamese Network (CNN-Siamese) and a multiscale Convolutional Siamese network (MCNN-Siamese). We have also explored and evaluated three pre-trained CNNs namely, ResNet-50, VGG-16 and NASNet-Large, and utilized them in Siamese network after appropriate modifications. In addition, a simple 5-layer neural network (sNN) is also utilised as Siamese subnetwork in the experiments. We have presented compelling experimental evidences in terms of correct classification accuracy (CCR) on CASIA-IrisV4 Twins’ dataset that manifest the effectiveness of ocular biometrics in identifying twins. The results achieve the existing state-of-the-art human level accuracy. Also, the proximity of CCRs of all the models asserts that the ocular regions in twins hold a significant correlation to label them as twins. We have also quantified the cross-domain capability of the proposed subnetworks (i.e. CNN and MCNN) on ND-GFI dataset that outperforms the state-of-the-art methods. Notably, for the ocular biometrics, to the best of our knowledge, there is currently no literature available as of date that explores the association between twins and subsequently unriddles the classification using Deep Learning.

Identifying twins based on ocular region features using deep representations

Blaming automated vehicles in difficult situations.

Bengali is the sixth most popular spoken language in the world. Computerized detection of handwritten Bengali (Bangla Lekha) character is very difficult due to the diversity and veracity of characters. In this paper, we have proposed a modified state-of-the-art deep learning to tackle the problem of Bengali handwritten character recognition. This method used the lesser number of iterations to train than other comparable methods. The transfer learning on Resnet-50 deep convolutional neural network model is used on pretrained ImageNet dataset. One cycle policy is modified with varying the input image sizes to ensure faster training. Proposed method executed on BanglaLekha-Isolated dataset for evaluation that consists of 84 classes (50 Basic, 10 Numerals and 24 Compound Characters). We have achieved 97.12% accuracy in just 47 epochs. Proposed method gives very good results in terms of epoch and accuracy compare to other recent methods by considering number of classes. Without ensembling, proposed solution achieves state-of-the-art result and shows the effectiveness of ResNet-50 for classification of Bangla HCR.

Bengali Handwritten Character Classification Using Transfer Learning on Deep Convolutional Network

Watson---Crick automata are finite automata working on double strands. Extensive research work has already been done on non-deterministic Watson---Crick automata and on deterministic Watson---Crick automata. State complexity of Watson---Crick automata has also been discussed. In this paper we analyze the state complexity of deterministic Watson---Crick automata with respect to non-deterministic block automata and non-deterministic finite automata. We also introduce new finite automata combining the properties of Watson---Crick automata and time varying automata. These automata have the unique property that the 1-limited stateless variant of these automata have the power to count. We further discuss the state complexity of time varying automata and time varying Watson---Crick automata.

State complexity of deterministic Watson---Crick automata and time varying Watson---Crick automata

In this paper, we propose a solution which uses state-of-the-art techniques in Deep Learning to tackle the problem of Bengali Handwritten Character Recognition ( HCR ). Our method uses lesser iterations to train than most other comparable methods. We employ Transfer Learning on ResNet 50, a state-of-the-art deep Convolutional Neural Network Model, pretrained on ImageNet dataset. We also use other techniques like a modified version of One Cycle Policy, varying the input image sizes etc. to ensure that our training occurs fast. We use the BanglaLekha-Isolated Dataset for evaluation of our technique which consists of 84 classes (50 Basic, 10 Numerals and 24 Compound Characters). We are able to achieve 96.12% accuracy in just 47 epochs on BanglaLekha-Isolated dataset. When comparing our method with that of other researchers, considering number of classes and without using Ensemble Learning, the proposed solution achieves state of the art result for Handwritten Bengali Character Recognition. Code and weight files are available at this https URL.

Bengali Handwritten Character Classification using Transfer Learning on Deep Convolutional Neural Network

MTSE U-Net: an architecture for segmentation, and prediction of fetal brain and gestational age from MRI of brain

1-way multihead quantum finite state automata (1QFA(k)) can be thought of modified version of 1-way quantum finite state automata (1QFA) and k-letter quantum finite state automata (k-letter QFA) respectively. It has been shown by Moore and Crutchfield as well as Konadacs and Watrous that 1QFA can’t accept all regular language. In this paper, we show different language recognizing capabilities of our model 1-way multihead QFAs. New results presented in this paper are the following ones: 1) We show that newly introduced 1-way 2-head quantum finite state automaton (1QFA(2)) structure can accept all unary regular languages. 2) A language which can’t be accepted by 1-way deterministic 2-head finite state automaton (1DFA((2)) can be accepted by 1QFA(2) with bounded error. 3) 1QFA(2) is more powerful than 1-way reversible 2-head finite state automaton (1RMFA(2)) with respect to recognition of language.

/pdf/1-way-multihead-quantum-finite-state-automata-x1c6z7q9vp.pdf

Debayan Ganguly

Papers

Bengali Handwritten Character Classification Using Transfer Learning on Deep Convolutional Network

State complexity of deterministic Watson---Crick automata and time varying Watson---Crick automata

Bengali Handwritten Character Classification using Transfer Learning on Deep Convolutional Neural Network

MTSE U-Net: an architecture for segmentation, and prediction of fetal brain and gestational age from MRI of brain

1-Way Multihead Quantum Finite State Automata