Handwritten Digit String Recognition Research Articles

Handwritten digit and Roman string recognition using gated CNN with federated learning

Handwritten digit and Roman string recognition using gated CNN with federated learning

On usage of an end-to-end deep neural architecture for handwritten digit string recognition

On usage of an end-to-end deep neural architecture for handwritten digit string recognition

DenseNet-CTC: An end-to-end RNN-free architecture for context-free string recognition

String recognition is one of the challenging tasks in document analysis and recognition areas. Recently, with the surge of interest in end-to-end segmentation-free methods, CRNN (Convolution Recurrent Neural Network), which is a combination of CNN (Convolutional Neural Network) and RNN-CTC (Recurrent Neural Network-Connectionist Temporal Classification), has been widely applied to string recognition. However, in some context-free cases, where a character is followed by arbitrary characters like the digit string, there may be no or very few context links in these strings. In this paper, we propose a new end-to-end RNN-free architecture especially for context-free string recognition and apply it to Handwritten Digit String Recognition (HDSR) task. The proposed architecture is based on CNN and CTC, but without the usage of RNN, and we apply column-wise fully connected layers to connect the convolutional layers and CTC directly. Moreover, to compensate for the possible reduction in modeling capabilities caused by the absence of RNN, we apply densely connected convolutional layers to extract efficient features. We test this new architecture on three public HDSR benchmarks (ORAND-CAR-A, ORAND-CAR-B and CVL HDS) and three other datasets that include a handwritten telephone/postcode dataset PhPAIS and two non-Arabic digit datasets (C-Bangla and C-Hindi). Furthermore, we generate three handwritten digit string datasets to further analyze the influence of RNN. The recognition results on all datasets demonstrate the superiority of the proposed model.

A comprehensive comparison of end-to-end approaches for handwritten digit string recognition

Over the last decades, most approaches proposed for handwritten digit string recognition (HDSR) have resorted to digit segmentation, which is dominated by heuristics, thereby imposing substantial constraints on the final performance. Few of them have been based on segmentation-free strategies where each pixel column has a potential cut location. Recently, segmentation-free strategies has added another perspective to the problem, leading to promising results. However, these strategies still show some limitations when dealing with a large number of touching digits. To bridge the resulting gap, in this paper, we hypothesize that a string of digits can be approached as a sequence of objects. We thus evaluate different end-to-end approaches to solve the HDSR problem, particularly in two verticals: those based on object-detection (e.g., Yolo and RetinaNet) and those based on sequence-to-sequence representation (CRNN).The main contribution of this work lies in its provision of a comprehensive comparison with a critical analysis of the above mentioned strategies on five benchmarks commonly used to assess HDSR, including the challenging Touching Pair dataset, NIST SD19, and two real-world datasets (CAR and CVL) proposed for the ICFHR 2014 competition on HDSR. Our results show that the Yolo model compares favorably against segmentation-free models with the advantage of having a shorter pipeline that minimizes the presence of heuristics-based models. It achieved a 97%, 96%, and 84% recognition rate on the NIST-SD19, CAR, and CVL datasets, respectively.

HDSR-Flor: A Robust End-to-End System to Solve the Handwritten Digit String Recognition Problem in Real Complex Scenarios

Automatic handwriting recognition systems are of interest for academic research fields and for commercial applications. Recent advances in deep learning techniques have shown dramatic improvement in relation to classic computer vision problems, especially in Handwritten Text Recognition (HTR). However, several approaches try to solve the problem of deep learning applied to Handwritten Digit String Recognition (HDSR), where it has to deal with the low number of trainable data, while learning to ignore any writing symbol around the digits (noise). In this context, we present a new optical model architecture (Gated-CNN-BGRU), based on HTR workflow, applied to HDSR. The International Conference on Frontiers of Handwriting Recognition (ICFHR) 2014 competition on HDSR were used as baselines to evaluate the effectiveness of our proposal, whose metrics, datasets and recognition methods were adopted for fair comparison. Furthermore, we also use a private dataset (Brazilian Bank Check - Courtesy Amount Recognition), and 11 different approaches from the state-of-the-art in HDSR, as well as 2 optical models from the state-of-the-art in HTR. Finally, the proposed optical model demonstrated robustness even with low data volume (126 trainable data, for example), surpassing the results of existing methods with an average precision of 96.50%, which is equivalent to an average percentage of improvement of 3.74 points compared to the state-of-the-art in HDSR. In addition, the result stands out in the competition’s CVL HDS set, where the proposed optical model achieved a precision of 93.54%, while the best result so far had been from Beijing group (from the competition itself), with 85.29%.

Unknown-Length Handwritten Numeral String Recognition Using Cascade of PCA-SVMNet Classifiers

Automatic recognition of handwritten digit string with unknown length has many potential real applications. The most challenging step in this problem is how to efficiently segment connected and/or overlapped digits exhibited in the input image. Most existing numeral string segmentation approaches combine several segmentation hypotheses to handle various types of connected digits. This paper proposes a new handwritten digit string recognition without applying any explicit segmentation techniques. The proposed method uses a new cascade of hybrid principal component analysis network (PCANet) and support vector machine (SVM) classifier called PCA-SVMNet. PCANet is an emerging unsupervised simple deep neural network typically with only two convolutional layers. The proposed PCA-SVMNet model adds a new fully connected layer trained separately using SVM optimization method. Cascaded stages of PCA-SVMNet classifiers are constructed and trained to recognize various types of isolated and connected digits. Every PCA-SVMNet classifier is trained separately using combinations of real and synthetic touching digits. The first 1D-PCA-SVMNet stage is trained to recognize isolated handwritten digits (0...9) while forwarding non-isolated digits to the next stages. Each of the following stages is designed to recognize a class of connected digits and forwards the higher class to its successor. Multiple stages can be added accordingly to classify more complex touching digits. The experimental results using NIST SD19 real dataset show that the cascade of PCA-SVMNet classifier efficiently recognizes unknown handwritten digit string without applying any sophisticated segmentation methods. The proposed method achieves state-of-the-art recognition accuracy compared to other segmentation-free techniques.

A Methodology for Feature Selection Using Multiobjective Genetic Algorithms for Handwritten Digit String Recognition

In this paper a methodology for feature selection for the handwritten digit string recognition is proposed. Its novelty lies in the use of a multiobjective genetic algorithm where sensitivity analysis and neural network are employed to allow the use of a representative database to evaluate fitness and the use of a validation database to identify the subsets of selected features that provide a good generalization. Some advantages of this approach include the ability to accommodate multiple criteria such as number of features and accuracy of the classifier, as well as the capacity to deal with huge databases in order to adequately represent the pattern recognition problem. Comprehensive experiments on the NIST SD19 demonstrate the feasibility of the proposed methodology.