SDSayan Das

Sayan Das

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Sayan Das has a total of 5 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Feature Ranking Using Novel Consistency Measure by Normalized Standard Deviation and Proposal of Three Novel Global Features for Online Signature Verification

    by Biswajit Kar, Bhimraj Prasai Chetry, Sayan Das

    Signature is a behavioral biometric that evolves throughout a person’s life. Feature extraction and ranking are very important steps towards online signature verification in order to achieve high efficiency. In our case, we have extracted 48 global features. A novel feature ranking technique based on consistency measure using normalized standard deviation is proposed here and is compared with well-established mRMR based ranking. The use of normalized standard deviation formula is the novel approach for feature ranking. Moreover, we have proposed three novel global features for consistency measure and mRMR based ranking. These three features has shown its importance in consistency based ranking as well as in mRMR based ranking. Consistency estimation of global features is important in one-class classification framework, where only the genuine signatures are available. All the features that are ranked by consistency measures and its weighting factors of feature vector are computed for every signer. More consistent features has given more weight for verification. Feature ranking shows the importance of each features for a particular system. Also it helps in feature selection to select the more discriminating features for a particular system and removing all irrelevant features. It saves the computational time and size of the model. Different global features has different scaling factor. Generally, normalization is done before ranking but in our technique we have ranked our features without normalization. Because normalization disrupt the statistical consistency. Therefore, we have used min-max normalization where all the features are converted to (0-1) range after ranking. In our proposed system, the proposed phase related global features shows more consistency in our ranking process. Similarly well-established mRMR feature ranking has also ranked our proposed novel features number 48 (Standard deviation of the phase), 43 (Entropy of shape signature function) and 47 (Mean of phase) within top 13 features. It is seen that both our proposed novel feature ranking technique based on consistency measure and well-established mRMR based ranking has shown almost similar performance. The proposed algorithms are verified with SVC 2004 database.


  • Open Access

    Fusion of Wavelet Features and Gabor Features for SVM-based Iris Verification

    by Sayan Das, Biswajit Kar

    Iris verification now become increasingly prominent in biometric-based person verification systems. It has gained a significant role in biometric systems due to its stability, high uniqueness, contactless and non-invasive properties. Iris has more inherent distinctive features than other biometrics. Feature extraction of iris plays a crucial role in this system for accurate person verification. Using the feature extraction process, unique iris features like textural patterns, crypts, and furrows of iris are extracted. In our study, we did a fusion of Discrete Wavelet Transform (DWT) features with multiple wavelet bases (db4, haar, coif3, and sym4) and Gabor features, which contain a good amount of textural and localized information. Fusion here indicates the concatenation of the extracted features using the above techniques. In this work, we studied this method on the full iris only so that a maximum number of features can be extracted. This combined approach yielded a significant 112 number of features. The extracted features are then verified using a support vector machine (SVM) classifier based on radial basis function (RBF) kernel with training vs testing split ratios of 8:2, 6:4, 4:6 and 2:8. In this study, we achieved the highest overall verification accuracy of 95.9% with training vs testing split ratio of 8:2. For other training vs testing split ratios of 6:4, 4:6 and 2:8 we achieved overall verification accuracies of 91.4%, 93.2% and 91.2% respectively. We got an overall verification accuracy of 92.9%, considering training vs testing ratios of 8:2, 6:4, 4:6 and 2:8.


  • Open Access

    Rotational Invariant Half Iris Based Person Verification Using Multi Wavelet and SVM

    by Sayan Das, Biswajit Kar

    Iris recognition is widely regarded as an efficient biometric technique due to the uniqueness and persistence of iris features. This recognition system has shown high accuracy and reliability compared to other biometric methods. However, extracting the relevant features from iris images poses a significant challenge. In this experimental research, it is found that fusing multiple wavelet features leads to improved accuracy, compared to using a single feature-based iris verification method. Specifically, support vector machines (SVM) as the classification algorithm is employed in this work. In this study, the accuracy of 19 types of different wavelets is assessed and the performance of the coiflet4 wavelet at the second-level horizontal component is specifically observed, which achieved an accuracy of 91.8%. On the other hand, the lowest accuracy was recorded for the bior1.3 wavelet at the fourth-level horizontal component, with a score of 65.6%. These findings emphasise the importance of selecting appropriate wavelets for achieving greater accuracy. This experimental setup yielded an accuracy of 98.62% for the CASIA database and 97.87% for the UBIRIS database. These accuracies indicate the effectiveness of utilizing multiple wavelet features and SVM-based classification in iris-based person verification. By exploiting the fusion of multiple wavelet characteristics, the recognition system’s ability to distinguish is enhanced, resulting in improved precision and dependability in the verification of individuals based on their iris patterns. The main work of this research is to evaluate the performance of half-iris-based verification. The goal is to address the challenges faced due to occlusion and rotational variations between the reference and test iris images. The analysis focuses on half iris verification with rotational variations of ±5◦,±10◦,±15◦ between the reference and test iris images. For the half iris, the strip size is reduced to 20x120 pixels, reflecting the analysis of only a portion of the iris and creating a more compact representation of its features. Multiple wavelet features and half iris-based person verification using SVM classifier techniques are applied here. The proposed technique for rotational invariant half-iris-based person verification achieves an accuracy of 98.59% for the CASIA database and 97.65% for the UBIRIS database. The half-iris-based verification results are very close to the full-iris-based verification results. The overall verification accuracy is far better compared to the reported results. The maximum verification accuracy reported for the CASIA database is 96.13%, whereas, we have achieved 98.59% accuracy for half iris with less number of features.

  • Open Access

    It Isn't Sh!tposting, It's My CAT Posting

    by Parthsarthi Rawat, Sayan Das, Jorge Aguirre, Akhil Daphara

    In this paper, we describe a novel architecture which can generate hilarious captions for a given input image. The architecture is split into two halves, i.e. image captioning and hilarious text conversion. The architecture starts with a pre-trained CNN model, VGG16 in this implementation, and applies attention LSTM on it to generate normal caption. These normal captions then are fed forward to our hilarious text conversion transformer which converts this text into something hilarious while maintaining the context of the input image. The architecture can also be split into two halves and only the seq2seq transformer can be used to generate hilarious caption by inputting a sentence.This paper aims to help everyday user to be more lazy and hilarious at the same time by generating captions using CATNet.