Browsing by Author "Jeevan, G."

Now showing 1 - 4 of 4

An empirical study of the impact of masks on face recognition
(Elsevier Ltd, 2022) Jeevan, G.; Zacharias, G.C.; Nair, M.S.; Rajan, J.
Face recognition has a wide range of applications like video surveillance, security, access control, etc. Over the past decade, the field of face recognition has matured and grown at par with the latest advancements in technology, particularly deep learning. Convolution Neural Networks have surpassed human accuracy in Face Recognition on popular evaluation tests such as LFW. However, most existing models evaluate their performance with an assumption of the availability of full facial information. The COVID-19 pandemic has laid forth challenges to this assumption, and to the performance of existing methods and leading-edge algorithms in the field of face recognition. This is in the wake of an explosive increase in the number of people wearing face masks. The reduced amount of facial information available to a recognition system from a masked face impacts their discrimination ability. In this context, we design and conduct a series of experiments comparing the masked face recognition performances of CNN architectures available in literature and exploring possible alterations in loss functions, architectures, and training methods that can enable existing methods to fully extract and leverage the limited facial information available in a masked face. We evaluate existing CNN-based face recognition systems for their performance against datasets composed entirely of masked faces, in contrast to the existing standard evaluations where masked or occluded faces are a rare occurrence. The study also presents evidence denoting an increased impact of network depth on performance compared to standard face recognition. Our observations indicate that substantial performance gains can be achieved by the introduction of masked faces in the training set. The study also inferred that various parameter settings determined suitable for standard face recognition are not ideal for masked face recognition. Through empirical analysis we derived new value recommendations for these parameters and settings. © 2021 Elsevier Ltd
Cross Task Temporal Consistency forÂ Semi-supervised Medical Image Segmentation
(Springer Science and Business Media Deutschland GmbH, 2022) Jeevan, G.; Pawan, S.J.; Rajan, J.
Semi-supervised deep learning for medical image segmentation is an intriguing area of research as far as the requirement for an adequate amount of labeled data is concerned. In this context, we propose Cross Task Temporal Consistency, a novel Semi-Supervised Learning framework that combines a self-ensembled learning strategy with cross-consistency constraints derived from the implicit perturbations between the incongruous tasks of multi-headed architectures. More specifically, the Signed Distance Map output of a teacher model is transformed to an approximate segmentation map which acts as a pseudo target for the student model. Simultaneously, the teacherâ€™s segmentation task output is utilized as the objective for the studentâ€™s Signed Distance Map derived segmentation output. Our proposed framework is intuitively simple and can be plugged into existing segmentation architectures with minimal computational overhead. Our work focuses on improving the segmentation performance in very low-labeled data proportions and has demonstrated marked superiority in performance and stability over existing SSL techniques, as evidenced through extensive evaluations on two standard datasets: ACDC and LA. Â© 2022, Springer Nature Switzerland AG.
Engagement Analysis of Students in Online Learning Environments
(Springer Science and Business Media Deutschland GmbH, 2022) Kamath, S.; Singhal, P.; Jeevan, G.; Annappa, B.
Engagement rate is considered a metric that measures the extent of engagement a particular content is receiving from the audience. In e-learning settings, educators want to observe the level of interest of learners to appropriately modify their courses and make the educational process more effective. In this paper, an ensemble approach is proposed to detect student engagement levels while watching an e-learning video. The ensemble model consists of a deep convolutional neural network (DCNN) for facial expression recognition and a deep recurrent neural network (DRNN) for establishing a relationship between eye-gaze and engagement intensity. OpenFace 2.0 toolbox abilities are leveraged for feature extraction. Experimental results on the test datasets give an accuracy of 55.64% on DAiSEE and an MSE of 0.0598 on Engagement in the Wild Dataset. Â© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
Semi-supervised structure attentive temporal mixup coherence for medical image segmentation
(Elsevier B.V., 2022) Pawan, S.J.; Jeevan, G.; Rajan, J.
Deep convolutional neural networks have shown eminent performance in medical image segmentation in supervised learning. However, this success is predicated on the availability of large volumes of pixel-level labeled data, making these approaches impractical when labeled data is scarce. On the other hand, semi-supervised learning utilizes pertinent information from unlabeled data along with minimal labeled data, alleviating the demand for labeled data. In this paper, we leverage the mixup-based risk minimization operator in a student–teacher-based semi-supervised paradigm along with structure-aware constraints to enforce consistency coherence among the student predictions for unlabeled samples and the teacher predictions for the corresponding mixup sample by significantly diminishing the need for labeled data. Besides, due to the intrinsic simplicity of the linear combination operation used for generating mixup samples, the proposed method stands at a computational advantage over existing consistency regularization-based SSL methods. We experimentally validate the performance of the proposed model on two public benchmark datasets, namely the Left Atrial (LA) and Automatic Cardiac Diagnosis Challenge (ACDC) datasets. Notably, on the LA dataset's lowest labeled data set-up (5%), the proposed method significantly improved the Dice Similarity Coefficient and the Jaccard Similarity Coefficient by 1.08% and 1.46%, respectively. Furthermore, we demonstrate the efficacy of the proposed method with a consistent improvement across various labeled data proportions on the aforementioned datasets. © 2022 Nalecz Institute of Biocybernetics and Biomedical Engineering of the Polish Academy of Sciences