Search

Existing Optical Mark Recognition (OMR) systems tend to be expensive and rigid in their operation, often resulting in erroneous evaluations due to strict correction protocols. This scenario airs the need for a flexible OMR system. Hence, in this work, we propose a lightweight transfer learning based Convolutional Neural Network (CNN) model, dubbed as OMRNet, which can classify answer boxes on any generalized OMR test sheet. Unlike most existing techniques that rely on image processing algorithms to recognize extracted answer boxes in two classes: confirmed and empty, the OMRNet is designed to classify the answer boxes into confirmed, crossed-out, and empty categories. That is, OMRNet is facilitating the crossing out of previously answered questions and thus removing the rigidity of templates in Multiple Choice Question (MCQ) tests. We have built OMRNet on top of a MobileNetV2 backbone connected to four fully connected layers with appropriate dropouts and activation functions in between. We have evaluated OMRNet on the Multiple Choice Answer Boxes dataset available at https://sites.google.com/view/mcq-dataset. We have performed experiments following a 5 fold cross validation scheme, and OMRNet has achieved accuracies of 95.29%, 95.88%, 93.97%, 97.45%, and 97.20%, with an average accuracy of 95.96%. Also, the experimental results confirm that the present model performs better than the compared state-of-the-art methods and standard CNN models in terms of accuracy, execution time, and memory required to store the trained module. Moreover, we have employed a quantization technique to make the trained module more memory efficient and deployed it to a web app using our own Representational State Transfer Application Programming Interface (REST API). It makes OMRNet available via a Hypertext Transfer Protocol (HTTP) endpoint, allowing potential users to connect to it via the Internet. The source code for the work is available at the following link: https://github.com/sa-y-an/OMRNet. [ABSTRACT FROM AUTHOR]

Video plays a key role in carrying authenticity, especially in the surveillance system, medical field, court evidence, journalism, and social media among others. However, nowadays the trust in videos is decreasing day by day due to the forgery of the videos made by easily accessible video editing tools. Hence, a thrust for finding a robust solution to the problem of video forgery detection arises. As a result, researchers around the world are indulging themselves to come up with various methods for the said problem. In this article, we have comprehensively discussed many such initiatives made by researchers across the globe, keeping the focus on recent trends. In addition to this, we have also covered a wide range of forgery detection techniques that follow either an active or a passive approach, while the state-of-the-art surveys made so far on this research topic include only a few specific cases. In this article, we have described some recent technologies that are used in video forging, made a summary of the performances (provided categorically) of all the techniques discussed here, and briefed the available datasets. Finally, we have concluded this survey by clearly mentioning some future directions of the video forgery detection research based on a thorough review of existing techniques. [ABSTRACT FROM AUTHOR]

Contrast enhancement is an important pre-processing task for several image and video processing applications. The objective of a contrast enhancement method is to improve the quality of the visual information contained in the images for further processing. Due to the enormous challenges, it is still considered as an open research problem. Several approaches have been proposed in the literature based on spatial and fr uency domain techniques. Among them, the Incomplete Beta Function (IBF) is a popular scheme used for image contrast enhancement. In the IBF based image contrast enhancement technique, quality of an image is improved by two controlling parameters. Under such conditions, these parameters need to be tuned for obtaining better outcomes. In this paper, a new gray-scale contrast enhancement algorithm is introduced where, instead of tuning the controlling parameters of IBF experimentally, their near-optimal values are calculated with the help of a recently published meta-heuristic algorithm called Artificial Electric Field Algorithm (AEFA). The proposed method has been compared with many state-of-the-art techniques in terms of some standard metrics considering three different datasets, namely Kodak, MIT-Adobe FiveK and USC-SIPI. Simulation results demonstrate that the proposed AEFA based image enhancement technique increases the overall image contrast and enriches the information present in the image. [ABSTRACT FROM AUTHOR]

The research community considers handwritten word recognition (HWR) as an open research problem to date. The reasons behind this are variations in intra-/interpersonal writing style, overlapping and/or touching characters in a word, degraded scanned document images, etc. Two major approaches, namely holistic and analytical, are followed by the researchers while designing an HWR system. In this work, we have followed the holistic approach as it works well on limited and pre-defined lexicon as compared to the analytical approach. As observed in the literature related to handwritten word recognition, irrespective of the approaches, researchers generally extract various local features from hypothetically partitioned segments of a word image while dealing with the said problem. However, no such work has been found which has considered inter-segment similarity that might carry some distinct information about different patterns (here, word segments). To this end, in the present work, we have used Hausdorff and Fréchet distances to quantize the similarity among all possible word segments taking two at a time. Along with this, conventional chain code histogram (a shape-based feature descriptor) and modified negative refraction-based shape transformation features have been used. Finally, a majority voting schema is used to combine outputs from six different classifiers. The model has been evaluated on two standard databases, namely IAM and IFN/ENIT, and the results obtained are promising in comparison with state-of-the-art holistic word recognition methods. Moreover, a performance comparison of the present method with some deep learning models confirms the usefulness of the proposed method. [ABSTRACT FROM AUTHOR]

In the recent past, video forgery has increased rapidly due to the easy availability of the required tools to accomplish that. Temporal copy-move or frame duplication is one of the most common video forgery methods in which a set of consecutive frames is copied somewhere in the same video. This work proposes an ensemble based method to detect duplicate frames from a video. In this method, first, the frames are pre-processed and then three different kinds of features - Haralick, custom Haralick and Local Binary Pattern (LBP) are extracted from each frame of a video. Next, lexicographical sorting is performed to arrange the frames having similar feature values consecutively. A filter method is applied to eliminate the false detection and it also creates a duplicate sequence for each category. In the end, a voting mechanism is employed to predict the duplicate frames, if they exist. The proposed method performs very well (detection accuracy is 99.32%, the true positive rate is 99.79%, and the true negative rate is 99.19%) over an in-house dataset containing 300 videos of seven different types. The robustness of the proposed method has also been tested by adding various postprocessing to the video samples. The proposed method outperforms many state-of-the-art methods used here for comparison. [ABSTRACT FROM AUTHOR]

Recent years have witnessed the surge of graph neural networks (GNNs) in various research fields. Compared to the convolutional neural networks (CNNs), GNNs happen to be more powerful and comprehensive as they use graphs to learn an embedding that contains information about the neighbourhood of a target node in a graph. Hence, GNN models have achieved exemplary performance in many tasks such as node classification, link prediction, and graph classification. In this work, we have come up with a solution to the problem of offline signature verification with the relational inductive biases of a GNN model. To apply the node classification strategy of a GNN model, a network is constructed. The network consists of training signature samples of a particular user, where each node represents a signature sample. GraphSAGE, an inductive representation learning algorithm, is utilized to train the model. Later, a test signature sample (i.e., a newly added node to the existing training graph) is classified into one of the two classes: genuine or forged using the trained model. The obtained results lay out a robust manifestation of the benefits of learned gray level co-occurrence matrix (GLCM) based features. The proposed method has been evaluated on two standard signature datasets, namely MCYT-75 and UTSig. The equal error rate (EER) values are 0.13 and 0.66 on MCYT-75 and UTSig datasets respectively. Use of a writer dependent schema with a distinctive model for each writer, and only with a handful number of training samples, the proposed model stands out amongst its predecessors. [ABSTRACT FROM AUTHOR]

In modern era it has become increasingly easier to manipulate and tamper digital images, one of the primary reasons being the boon of commonplace availability of powerful image editing tools and software. These tools become a bane when used for malicious reasons as users can possibly add or remove important features from an image without leaving any obvious marks of tampering. Hence the need of forgery detection techniques which show high accuracy in detection arises. One of the most prevalent forms of image tampering is the copy-move forgery attack. In this type of forgery, a part of an image is copied and then pasted somewhere else in the same image with the intent to hide key features of the image. This paper introduces a new copy-move image forgery detection technique which relies on a texture feature descriptor called Local Tetra Pattern (LTrP) for block level image comparison used to localize tampered region(s). Initially, the input image is divided into overlapping blocks, then LTrP features are extracted from each block to form a single feature vector. Next, the feature vectors of all image blocks are sorted lexicographically, and then similar blocks are identified by matching the features from neighboring blocks. Finally, blocks matched falsely due to the presence of homogeneous color information like sea, field, and sky are removed using a shift vector aided outlier removal method. Experiments have been conducted on two standard datasets - GRIP and CoMoFoD. We have obtained 0.9834 and 0.9093 average F1 scores at pixel-level for GRIP and CoMoFoD datasets respectively. The experimental results demonstrate that the proposed technique has been able to detect the forged regions with higher accuracy as compared to many state-of-the-art copy-move forgery detection methods. Moreover, experimental results on CoMoFoD dataset show that the method is able to correctly detect the forgery even after various post-processing attacks. The source code of proposed method is available at https://github.com/RollingThunderSagnik/LTrP-Copy-Move-Forgery-Detection. [ABSTRACT FROM AUTHOR]

Digital face manipulation has become a concern in the last few years due to its harmful impacts on society. It is especially concerning for high-profile celebrities because their identities can be easily manipulated using mobile or web applications such as FaceSwap and FaceApp. These manipulated faces are so close to real ones that it becomes really hard to detect them, even with bare eyes. Though deep learning-based models are predominantly used by researchers, they hardly check for the presence of irrelevant or redundant features produced by those models. To this end, we have proposed a hierarchical feature selection (HFS)-based method to detect deepfake images or videos. First, we have extracted both handcrafted and deep learning features from the inputs. Next, the HFS is applied to select a near optimal set of features. In each stage of it, a hybrid feature selection method is employed that integrates a population-based meta-heuristic model, called Grey Wolf Optimization, and a single solution-based meta-heuristic model, called the Vortex Search algorithm. We have evaluated our model on three publicly available datasets, namely Celeb-DF (V2), FaceForensics++, and Deepfake Detection Challenge (DFDC). The model provides 99.35%, 99.16%, and 85.67% AUC scores on the Celeb-DF (V2), FaceForensics++, and DFDC datasets, respectively, while using only 11.50%, 12.65%, and 10.22% of actual features. Besides, our model outperforms most of the state-of-the-art methods found in our literature review and evaluated on these three datasets. [ABSTRACT FROM AUTHOR]

The prime goal of creating synthetic digital data is to generate something very closer to real ones when the original data are scarce. However, the trustworthiness of such digital content is dipping potentially in society owing to malicious users. Deepfake method that uses computer graphics and computer vision techniques to replace the face of one person with the face of a different person is becoming an area of big concern. Such techniques can easily be used to hide the identity of a person. Therefore, a method is needed to verify the originality of such face images/videos. To this end, we design a deep learning model enhanced with visual attention technique to differentiate manipulated videos/images (generated by deepfake methods) from real ones. At first, we extract the face region from video frames and then pass the same through the pre-trained Xception model to obtain the feature maps. Next, with the help of the visual attention mechanism, we mainly try to focus on the deepfake video manipulation leftover artifacts. We evaluate our model on two publicly available datasets, namely FaceForensics++ and Celeb-DF (V2), and our model outperforms many state-of-the-art methods tested on these two datasets. Source code of the proposed method can be found at: https://github.com/tre3x/Deepfake-Video-Forgery-Detection. [ABSTRACT FROM AUTHOR]

Disease prediction from diagnostic reports and pathological images using artificial intelligence (AI) and machine learning (ML) is one of the fastest emerging applications in recent days. Researchers are striving to achieve near-perfect results using advanced hardware technologies in amalgamation with AI and ML based approaches. As a result, a large number of AI and ML based methods are found in the literature. A systematic survey describing the state-of-the-art disease prediction methods, specifically chronic disease prediction algorithms, will provide a clear idea about the recent models developed in this field. This will also help the researchers to identify the research gaps present there. To this end, this paper looks over the approaches in the literature designed for predicting chronic diseases like Breast Cancer, Lung Cancer, Leukemia, Heart Disease, Diabetes, Chronic Kidney Disease and Liver Disease. The advantages and disadvantages of various techniques are thoroughly explained. This paper also presents a detailed performance comparison of different methods. Finally, it concludes the survey by highlighting some future research directions in this field that can be addressed through the forthcoming research attempts. [ABSTRACT FROM AUTHOR]

The novel coronavirus (COVID-19), has undoubtedly imprinted our lives with its deadly impact. Early testing with isolation of the individual is the best possible way to curb the spread of this deadly virus. Computer aided diagnosis (CAD) provides an alternative and cheap option for screening of the said virus. In this paper, we propose a convolution neural network (CNN)-based CAD method for COVID-19 and pneumonia detection from chest X-ray images. We consider three input types for three identical base classifiers. To capture maximum possible complementary features, we consider the original RGB image, Red channel image and the original image stacked with Robert's edge information. After that we develop an ensemble strategy based on the technique for order preference by similarity to an ideal solution (TOPSIS) to aggregate the outcomes of base classifiers. The overall framework, called TOPCONet, is very light in comparison with standard CNN models in terms of the number of trainable parameters required. TOPCONet achieves state-of-the-art results when evaluated on the three publicly available datasets: (1) IEEE COVID-19 dataset + Kaggle Pneumonia Dataset, (2) Kaggle Radiography dataset and (3) COVIDx. [ABSTRACT FROM AUTHOR]

Outlier or anomaly detection is the process through which datum/data with different properties from the rest of the data is/are identified. Their importance lies in their use in various domains such as fraud detection, network intrusion detection, and spam filtering. In this paper, we introduce a new outlier detection algorithm based on an ensemble method and distance-based data filtering with an iterative approach to detect outliers in unlabeled data. The ensemble method is used to cluster the unlabeled data and to filter out potential isolated outliers from the same by iteratively using a cluster membership threshold until the Dunn index score for clustering is maximized. The distance-based data filtering, on the other hand, removes the potential outlier clusters from the post-clustered data based on a distance threshold using the Euclidean distance measure of each data point from the majority cluster as the filtering factor. The performance of our algorithm is evaluated by applying it to 10 real-world machine learning datasets. Finally, we compare the results of our algorithm to various supervised and unsupervised outlier detection algorithms using Precision@n and F-score evaluation metrics. [ABSTRACT FROM AUTHOR]

In ancient times, there was no system to record or document music. A basic notation system to write European music was formulated around 14th century in the Baroque period which slowly evolved into the standard notation system that we have today. Later, the musical pieces from the classical and post-classical period of European music were documented as scores using this standard European staff notations. These notations are used by most of the modern genres of music due to their versatility. Hence, it is very important to develop a method that can store such music sheets containing handwritten music scores digitally. Optical music recognition (OMR) is a system that automatically interprets the scanned handwritten music scores. In this work, we have proposed a classifier ensemble of deep transfer learning models with support vector machine (SVM) as the aggregator for handwritten music symbol recognition. We have applied three pre-trained deep learning models, namely ResNet50, GoogleNet and DenseNet161 (each trained on ImageNet), and fine-tuned on our target datasets i.e., music symbol image datasets. The proposed ensemble technique can capture a more complex association of the base classifiers, thus improving the overall performance. We have evaluated the proposed model on five publicly available standard datasets, namely Handwritten Online Music Symbols (HOMUS), Capitan_Score_Uniform, Capitan_Score_Non-uniform, Rebelo_real and Fornés, and achieved state-of-the-art results for all these datasets. Additionally, we have evaluated our model on publicly available two non-music symbols datasets, namely CMATERdb 2.1.2 containing 120 handwritten Bangla city names and CMATERdb 3.1.1 dataset containing handwritten Bangla numerals to validate its effectiveness on diversified datasets. The source code of this present work is available at https://github.com/ashis0013/Music-Symbol-Recognition. [ABSTRACT FROM AUTHOR]

In predictive modelling it is important to use any feature selection methods as irrelevant features when used with powerful classifiers can lead to over-fitting and thus create models which fail to perform as good as when these features are not used. Particularly it is important in case of disease datasets where various features or attributes are available through the patients' medical records and many features in these datasets may not be relevant to the diagnosis of some specific disease. Wrong models in this case can be disastrous and lead to wrong diagnosis, and maybe in extreme cases lead to loss of life. To this end, we have used a wrapper based feature selection model for the said purpose. In recent years, Grasshopper Optimization Algorithm (GOA) has proved its superiority over other optimization algorithms in different research areas. In this paper, we propose an improved version of GOA, called (LAGOA), which uses Learning Automata (LA) for adjusting the parameters of GOA in an adaptive way, and two-phase mutation for enhancing exploitation capability of the algorithm. LA is used for adjusting the parameter values of each grasshopper in the population individually. In two-phase mutation the first phase reduces the number of selected features while preserving high classification accuracy, while the second phase adds relevant features which increase the classification accuracy. Proposed method has been applied to Breast Cancer (Wisconsin), Breast Cancer (Diagnosis), Statlog (Heart), Lung Cancer, SpectF Heart and Hepatitis datasets taken from UCI Machine Learning Repository. Experimental results confirm its superiority over state-of-the-art methods considered here for comparison. [ABSTRACT FROM AUTHOR]

Breast cancer, the most common invasive cancer, causes deaths of thousands of women in the world every year. Early detection of the same is a remedy to lessen the death rate. Hence, screening of breast cancer in its early stage is utmost required. However, in the developing nations not many can afford the screening and detection procedures owing to its cost. Hence, an effective and less expensive way of detecting breast cancer is performed using thermography which, unlike other methods, can be used on women of various ages. To this end, we propose a computer aided breast cancer detection system that accepts thermal breast images to detect the same. Here, we use the pre-trained DenseNet121 model as a feature extractor to build a classifier for the said purpose. Before extracting features, we work on the original thermal breast images to get outputs using two edge detectors - Prewitt and Roberts. These two edge-maps along with the original image make the input to the DenseNet121 model as a 3-channel image. The thermal breast image dataset namely, Database for Mastology Research (DMR-IR) is used to evaluate performance of our model. We achieve the highest classification accuracy of 98.80% on the said database, which outperforms many state-of-the-art methods, thereby confirming the superiority of the proposed model. Source code of this work is available here: https://github.com/subro608/thermogram_breast_cancer- [ABSTRACT FROM AUTHOR]

Handwriting is used to distribute information among people. To access this information for further analysis the page needs to be optically scanned and converted to machine recognizable form. Due to unconstrained writing styles along with connected and overlapping characters, handwriting recognition remains a challenging task. Most of the methods in the literature use lexicon-based approaches and train their models on large datasets having near 50 K word samples to achieve good results. This results in high computational requirements. While these models use around 50 K words in their dictionary when recognizing handwritten English text, the actual number of words in the dictionary is much higher than this. To this end, we propose a handwriting recognition technique to recognize handwritten English text based on a YOLOv3 object recognition model that is lexicon-free and that performs sequential character detection and identification with a low number of training samples (only 1200 word images). This model works well without any dependency on writers' style of writing. This is tested on the IAM dataset and it is able to achieve 29.21% Word Error Rate and 9.53% Character Error Rate without a predefined vocabulary, which is on par with the state-of-the-art lexicon-based word recognition models. [ABSTRACT FROM AUTHOR]

Biometric verification systems are used to recognize people based on their uniqueness or characteristics. Signature is considered as one of the most commonly used biometric that individualizes a human being. It is generally used to keep individual's privacy in many places such as banking sectors, academic institutes, office premises and trading. But increase of criminal attempts in falsifying an individual's signature, known as signature forgeries, motivates the researchers to develop computerized systems that can verify the genuineness of a questioned signature. Though many researches have been performed till date, but the issue of identifying skilled forgeries still remains a major concern for the researchers. To this end, in this work, we have designed a two-tier ensemble based writer dependent and language- invariant online signature verification system. In doing so, we have first extracted three different categories of features from each input signature: physical, frequency based and statistical, and then designed a feature-classifier based ensemble (i.e., Ensemble#1) using seven different classifiers. The predictions obtained from the seven classifiers are combined using normalised distribution summation strategy. Decisions obtained from Ensemble#1 are then fed to Ensemble#2, where a majority voting based approach is followed, to identify the input signature as genuine or forged. Our system is evaluated on two standard datasets: SVC 2004 (Task-II) and MCYT-100 in a writer dependent way. The equal error rate (ERR) and accuracy on SVC 2004 dataset are 2.20 and 98.43% respectively, and on MCYT-100 dataset these are 2.84 and 97.87% respectively. The GAR@0.01FAR value obtained for the SVC-2004 dataset is 94.50% while it is 92.90% for MCYT-100 dataset. We have also compared our results with some state-of-the-art methods, and it has been found that our method performs better than most of these methods. The code of this work is available at: https://github.com/prat1999/Online_Signature_Verification. [ABSTRACT FROM AUTHOR]

Though handwriting recognition is a well-explored research area for decades, there are a few sub-areas of this field that have still not obtained much attention from the researchers. Some examples include recognition of hand-drawn graphics components like circuit components and diagrams. Complete digitization of such handwritten documents is not possible without automatic conversion of the said circuit diagrams. Besides, to date, in most of the cases for commercial circuit design purposes, concerned people manually enter the components into the simulating software like Cadence, Spice to analyze the circuit and judge its performance. In this work, it has been tried to move one step towards automating this process by recognizing the hand-drawn circuit components which are considered as the most important step for this automation. The present endeavour is to design a two-stage convolutional neural network (CNN)-based model that recognizes the hand-drawn circuit components. In the first stage, all the similar-looking (i.e., similar shape and structure) circuit components are clustered into a single group using visual perception and input from confusion matrix of single-stage CNN-based classification, and in the later stage, the circuit components belong to the same group are classified into their actual classes. The proposed model has been evaluated on a self-made database where 20 different classes of hand-drawn circuit components are considered. The experimental outcome shows that the proposed two-stage classification model provides an accuracy 97.33% which is much higher than the single-stage method, which provides an accuracy of 86.00%. [ABSTRACT FROM AUTHOR]

Many real-life datasets are imbalanced in nature, which implies that the number of samples present in one class (minority class) is exceptionally less compared to the number of samples found in the other class (majority class). Hence, if we directly fit these datasets to a standard classifier for training, then it often overlooks the minority class samples while estimating class separating hyperplane(s) and as a result of that it missclassifies the minority class samples. To solve this problem, over the years, many researchers have followed different approaches. However the selection of the true representative samples from the majority class is still considered as an open research problem. A better solution for this problem would be helpful in many applications like fraud detection, disease prediction and text classification. Also, the recent studies show that it needs not only analyzing disproportion between classes, but also other difficulties rooted in the nature of different data and thereby it needs more flexible, self-adaptable, computationally efficient and real-time method for selection of majority class samples without loosing much of important data from it. Keeping this fact in mind, we have proposed a hybrid model constituting Particle Swarm Optimization (PSO), a popular swarm intelligence-based meta-heuristic algorithm, and Ring Theory (RT)-based Evolutionary Algorithm (RTEA), a recently proposed physics-based meta-heuristic algorithm. We have named the algorithm as RT-based PSO or in short RTPSO. RTPSO can select the most representative samples from the majority class as it takes advantage of the efficient exploration and the exploitation phases of its parent algorithms for strengthening the search process. We have used AdaBoost classifier to observe the final classification results of our model. The effectiveness of our proposed method has been evaluated on 15 standard real-life datasets having low to extreme imbalance ratio. The performance of the RTPSO has been compared with PSO, RTEA and other standard undersampling methods. The obtained results demonstrate the superiority of RTPSO over state-of-the-art class imbalance problem-solvers considered here for comparison. The source code of this work is available in https://github.com/Sayansurya/RTPSO_Class_imbalance. [ABSTRACT FROM AUTHOR]

Word spotting in handwritten document images is a field of immense interest due to its widespread applications. Recognition-free and recognition-based approaches are the two comprehensively studied regimes for the said problem out of which the first one is more realistic for practical applications. In literature, several works have been found that have used contour and distance-based measures for matching of the profiles of two word images. Although this is a prudent choice for printed words, the same often faces bottlenecks for unconstrained handwriting. To this end, this work applies dynamic time warping algorithm on logarithmic profiles of handwritten word images to lessen the uncontrolled profile variation that occurs due to elongation while writing some characters. We have considered both global and local interpretations of a word image by dividing it vertically into a number of sub-parts. This multi-view analysis provides close-up views of different approximations for the same word image. Finally, a voting scheme is evoked to produce the final decision. Besides, we have adopted a pruning method to pre-filter the target word images prior to applying the voting-based word matching scheme. The method has been tested on word images, taken from Qatar University Writer Identification database. We have obtained satisfactory results as compared to many state-of-the-art methods that also include deep learning-based feature extraction models. [ABSTRACT FROM AUTHOR]

With a wide variety of forms being generated in different organizations daily, efficient and quick retrieval of information from these forms becomes a pressing need. The data on these forms are imperative to any commercial or professional purpose and thus, efficient retrieval of this data is important for further processing of the same. An automatic form processing system retrieves the content of a filled-in form image for useful storage of the same. Despite a large population of the world speaking in Bangla, to the best of our knowledge, there is no significant research work found in literature which deals with form data written in Bangla. To bridge this research gap, in the present scope of the work, we have developed a system that addresses four important aspects of processing of form data written using Bangla script. Our work has primarily been divided into four major modules: touching component separation, text non-text separation, handwritten printed text separation and alphabet numeral separation. The vital problem of touching component separation has been addressed using a novel rule-based method. For text non-text separation, handwritten printed text separation and alphabet numeral separation, we have used a machine learning based approach using feature engineering where the model for each case has been finalized after exhaustive experiments. Further, in each of the last three modules, we have applied some new features along with some existing features to appropriately tune the modules to obtain optimum results. Notably, we have also prepared a self-made database of filled-in forms. To create different training models, first the filled-in form images are binarized, and then different types of components are colored uniquely to obtain images which act as the ground truth for our reference. Evaluation of modules on the said database produces reasonably satisfactory results considering the complexity of the research problem. The code along with some filled-in sample form images and their respective ground truth images are provided in the link https://github.com/rajdeep-cse17/Form_Processing. [ABSTRACT FROM AUTHOR]

Recognition of unconstrained handwritten word images is an interesting research problem which gets more challenging when lexicon-free words are considered. Prerequisite for developing a lexicon-free handwritten word recognition technique is the segmentation of a word image into its constituent character set. Therefore, a competent character segmentation technique is required to design a comprehensive word recognition module. However, the literature study reveals that there is no standard word image database with ground truth information. As a result, most character segmentation algorithms found in the literature rely on self-made databases with manual evaluation. To fill the research need, in the present scope of the work, a comprehensive database consisting of handwritten Bangla word images is prepared primarily for evaluating any character segmentation algorithms. Additionally, the present work also provides two types of ground truth images related to segmented character shapes of the word images. Besides, an evaluation tool is developed for assessing the performance of any character segmentation algorithm on the developed benchmark database. The benchmark result, as found here, is 0.9212 (F-score) which outperforms some state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Handwritten music symbol recognition is considered by the research fraternity as a critical research problem. It becomes more critical when the symbols are collected from handwritten music sheets in offline mode. Most of the research findings, available in the literature, have tried to recognize the said symbols using various shape based features. But this approach limits system performance when we dealt with lookalike symbols such as half note, eight note and quarter note. To encounter this, in the present work we have used a texture based feature descriptor, called Daisy, for the said purpose. Though Daisy descriptor yields reasonably good recognition accuracy, but it generates a high dimensional feature vector. Hence, in this work, Quantum concept inspired Grey Wolf Optimization, named as QGWO, has been applied to select optimal feature subset from this high dimensional feature vector. We have applied the proposed method on six different standard music symbol datasets that include HOMUS, Capitan_score_uniform, Capitan_score_non-uniform, Fornés, Rebelo_real and Rebelo_synthetic datasets. On these datasets we have achieved recognition accuracies 93.07%, 99.22%, 99.20%, 99.49% and 100.00% respectively with 39.63%, 49.75%, 42.50%, 67.62%, 54.37% and 71.25% of actual feature dimension (i.e., 800) respectively. Additionally, we have compared our results with some state-of-the-art methods along with two recent deep learning based models, and it has been found that the present approach outperforms those. [ABSTRACT FROM AUTHOR]

Circuit diagram is the very foundation of electrical and electronic sciences. A circuit diagram consists of various symbols called circuit components that specify the functionality of that circuit. Every day-to-day gadgets that we use are made up with a number of electrical/electronic circuits to play out their particular tasks. Till date circuit designers have to physically enter all data from the hand-drawn circuits into computers, and this procedure requires some investment in terms of time and carries mistakes with high likelihood. To this end, in this paper, we propose a method that relaxes this constraint by introducing a method for recognition of hand-drawn electrical and electronic circuit components, with both analog and digital components included. In the proposed method, the pre-processed images of circuit components are used for training and testing a recognition model using a feature set consisting of a texture based feature descriptor, called histogram of oriented gradients (HOG), and shape based features that include centroid distance, tangent angle, and chain code histogram. In addition, the texture based feature, being large in number compared to others is optimized using a feature selection algorithm called ReliefF. Classification of components is done by using sequential minimal optimization (SMO) classifier. The proposed method has been evaluated on a dataset of 20 different circuit components with 150 samples in each class. The experimental outcome shows that the proposed approach provides average 93.83% accuracy on the present database. We also compare our method with some of the state-of-the-art methods and we see that our method outperforms these methods. [ABSTRACT FROM AUTHOR]

Handwritten word recognition, a classical pattern recognition problem, converts a word image into its machine editable form. Mainly two basic approaches are followed to solve this problem, one is segmentation-based and the other is holistic. A number of research attempts have shown that the holistic approach performs better than its counterpart when the lexicon is predefined, fixed and small in size. Relying on this, initial benchmark recognition accuracy on CMATERdb2.1.2, a publicly available database consists of handwritten city names in Bangla, was reported following a holistic word recognition protocol. In the present work, we have followed the same trend to recognize the word samples of the said database and set a new benchmark recognition accuracy. A sparse convolutional neural network (CNN)-based model which is a low-cost trainable model has been developed for this. We have relied on a recently proposed hypothesis, known as lottery ticket hypothesis for pruning the layers of CNN model methodically, and derived a low-resource model having much less number of training parameters. This model competently surpasses the previously reported recognition accuracy on the said database by a significant margin with an axed training cost. [ABSTRACT FROM AUTHOR]

Feature selection plays a key role in reducing the dimensionality of a feature vector by discarding redundant and irrelevant ones. In this paper, a Genetic Algorithm-based hierarchical feature selection (HFS) model has been designed to optimize the local and global features extracted from each of the handwritten word images under consideration. In this context, two recently developed feature descriptors based on shape and texture of the word images have been taken into account. Experimentation is conducted on an in-house dataset of 12,000 handwritten word samples written in Bangla script. This database comprises names of 80 popular cities of West Bengal, a state of India. Proposed model not only reduces the feature dimension by nearly 28%, but also enhances the performance of the handwritten word recognition (HWR) technique by 1.28% over the recognition performance obtained with unreduced feature set. Moreover, the proposed HFS-based HWR system performs better in comparison with some recently developed methods on the present dataset. [ABSTRACT FROM AUTHOR]

Due to the cursive nature, segmentation of handwritten Bangla words into characters and also recognition of the same sometimes become a very challenging problem to the researchers. Presence of comparatively large character set along with modifiers, ascendants, descendants, and compound characters makes the segmentation task more complex. As holistic method avoids such character-level segmentation, it is generally useful for the recognition of words written in any such complex scripts. In the present work, a holistic handwritten word recognition method is developed using a feature descriptor, designed by combining different Elliptical, Tetragonal and Vertical pixel density histogram-based features. Recognition process is carried out separately using two classifiers, namely multi-layer perceptron (MLP) and support vector machine (SVM). For evaluation of the proposed method, a database of 18,000 handwritten Bangla word images, having 120 word classes, is prepared. The proposed system performs comparatively better with SVM than MLP for the prepared dataset. It has achieved 83.64% accuracy at best case and 79.38% accuracy on an average using fivefold cross-validation. The current method has also outperformed some recently reported holistic word recognition technique tested on the developed dataset. In addition to that the database, prepared in this work, is made freely available to fill the absence of a publicly available standard database for holistic Bangla word recognition. [ABSTRACT FROM AUTHOR]