Article
Computer Science, Information Systems
P. K. Athira, C. J. Sruthi, A. Lijiya
Summary: With a large population of hearing impaired and vocal disabled individuals in India, the development of a sign language interpretation system has become highly important. This paper proposes a novel vision-based gesture recognition system that can recognize Indian Sign Language gestures and finger spelling words from live video. The system achieved high accuracy in recognizing finger spelling alphabets and single-handed dynamic words.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2022)
Article
Computer Science, Information Systems
Soumen Das, Saroj Kr Biswas, Biswajit Purkayastha
Summary: The deaf community faces challenges in communication with the hearing community. Traditional methods like employing sign language interpreters are not efficient and cost-effective. This paper proposes an automated sign language recognition system (AISLRSEW) that combines CNN and local handcrafted features to improve recognition accuracy, specifically in emergency situations with ISL words. Through evaluation and comparison, the proposed model achieves an average accuracy of 94.42%, outperforming existing models.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Multidisciplinary Sciences
Medet Mukushev, Aidyn Ubingazhibov, Aigerim Kydyrbekova, Alfarabi Imashev, Vadim Kimmelman, Anara Sandygulova
Summary: This paper presents a new large-scale signer independent dataset for Kazakh-Russian Sign Language, which can be used for performance evaluations of Continuous Sign Language Recognition and Translation tasks. The dataset contains a high degree of linguistic and inter-signer variability and is publicly available.
Article
Computer Science, Information Systems
Sunusi Bala Abdullahi, Kosin Chamnongthai
Summary: This article proposes a new S-p model called IDF-Sign, which is built using a spatial-temporal multivariate pairwise consistency feature ranking approach. It solves the problem of inconsistent hand and body features in sign language recognition and translation. Experimental results demonstrate that IDF-Sign achieves high recognition performance on various datasets.
Article
Computer Science, Artificial Intelligence
Zekeriya Katilmis, Cihan Karakuzu
Summary: Sign language is vital for the communication of hearing impaired individuals. This study focuses on recognizing two-handed dynamic words in Turkish Sign Language (TSL) using the LMC device. By repeating 26 dynamic words selected based on their similarities and differences, two data sets were extracted. A three-stage strategy comprising of data regularization, feature selection, and dimension reduction was applied to these feature sets to present word recognition performances from various aspects. The recognition performance was evaluated using six different ELM networks, and the most effective Meta-ELM classifier was identified.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Sakshi Sharma, Sukhwinder Singh
Summary: Hand gestures are crucial for communication and form the foundation of sign language, which is a visual form of communication. A deep learning CNN model designed for recognizing gesture-based sign language achieved high classification accuracy with fewer model parameters. The proposed model outperformed existing techniques in classifying gestures accurately with minimal error rates.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Information Systems
Sameena Javaid, Safdar Rizvi
Summary: Sign language is used to bridge the communication gap for people with hearing and speaking impairments. This study proposes a novel system, called Sign Language Action Transformer Network (SLATN), which localizes hand, body, and facial gestures in video sequences using a Transformer-style network. The model achieves faster speed and higher accuracy compared to traditional activity recognition methods.
CMC-COMPUTERS MATERIALS & CONTINUA
(2023)
Article
Engineering, Multidisciplinary
P. Jayanthi, Ponsy R. K. Sathia Bhama, B. Madhubalasri
Summary: Sign Language Recognition (SLR) aims to interpret signs for effective communication between hearing or speaking disabled people and normal people. The key to recognizing signs is the scarce key information about the gestures. Deep Convolutional Neural Network is used to identify gestures from video, while Recurrent Neural Network-Long Short-Term Memory verifies the semantics of the gesture sequence and converts it into speech. A model is proposed to construct meaningful sentences from continuous gestures, by processing only the principal elements for classification. The sentences are converted into voiceover for elegant communication between impaired and normal people. The model achieved high accuracy in gesture detection and word prediction using neural networks, making communication harmonious.
JOURNAL OF SCIENTIFIC & INDUSTRIAL RESEARCH
(2023)
Article
Computer Science, Artificial Intelligence
Zhengzhe Liu, Lei Pang, Xiaojuan Qi
Summary: This study introduces a mutual enhancement network (MEN) for joint sign language recognition and education, which formulates the sign language recognition system and the sign language education system as an estimation-maximization (EM) framework to boost performance. Experimental results validate the superiority of the proposed framework.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Computer Science, Information Systems
Khadidja Sadeddine, Zohra Fatma Chelali, Rachida Djeradi, Amar Djeradi, Sidahmed Benabderrahmane
Summary: This paper proposes a static hand gesture recognition approach based on image descriptors, utilizing Principal Component Analysis to reduce dimensionality, and tested with neural network classifiers, K-Nearest Neighbor classifiers, and combined classifiers, achieving high accuracy despite challenging data acquisition conditions.
JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION
(2021)
Article
Computer Science, Information Systems
Yusen Wang, Fan Li, Yadong Xie, Chunhui Duan, Yu Wang
Summary: In this article, an end-to-end American sign language (SLR) system is proposed for both word-level and sentence-level recognition. The system utilizes inaudible acoustic signals to estimate channel information and capture sign language in real time. Channel impulse response is used to represent each sign language gesture, enabling finger-level recognition. The system also considers conversion movements between two words as additional labels for training the sentence-level classification model. A prototype system is implemented and experimental results show promising performance, achieving an accuracy of 97.2% at word-level recognition and a word error rate of 0.9% at sentence-level recognition.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Automation & Control Systems
Pedro M. Ferreira, Diogo Pernes, Ana Rebelo, Jaime S. Cardoso
Summary: Sign language recognition (SLR) is a key technology in human-computer interaction, but the large intersigner variability in sign languages poses a challenge. To address this, researchers propose a novel deep neural network model that learns discriminative signer-independent latent representations from input data. This model aims to create a truly signer-independent system that is robust to different test signers, as demonstrated by experimental results in various SLR databases.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2021)
Article
Computer Science, Information Systems
Hezhen Hu, Junfu Pu, Wengang Zhou, Houqiang Li
Summary: This article presents a unified framework for multilingual continuous sign language recognition, which improves model performance by sharing a visual encoder and introducing language embeddings. Experimental results show that this method outperforms individually trained models and other state-of-the-art algorithms.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Psychology, Experimental
Miriam A. Novack, Diane Brentari, Susan Goldin-Meadow, Sandra Waxman
Summary: The research found that the early link between language and cognition is not limited to spoken language, but sign language can also help infants form object categories and further cognitive development. However, compared to sign language, hearing infants' responses to their native language (spoken language) were more positive.
Article
Computer Science, Information Systems
Hamzah Luqman
Summary: Sign language recognition is an important research area, and this paper proposes a trainable deep learning network that effectively captures the spatiotemporal information of isolated sign language. The network consists of three modules and incorporates techniques to handle variations in sign samples. Experimental results show that this approach outperforms other techniques in recognizing static signs and achieves superior performance on various sign language datasets.
Article
Engineering, Electrical & Electronic
R. Elakkiya, G. Kavithaa, Vahid Samavatian, K. Alhaifi, Alireza Kokabi, Hossein Moayedi
IEEE TRANSACTIONS ON POWER ELECTRONICS
(2020)
Review
Computer Science, Artificial Intelligence
R. Elakkiya
Summary: The research examines the impact of machine learning in sign language recognition, highlighting the challenges faced by current systems and potential solutions. It compares various approaches and emphasizes multilingual sign recognition.
JOURNAL OF AMBIENT INTELLIGENCE AND HUMANIZED COMPUTING
(2021)
Article
Computer Science, Artificial Intelligence
R. Elakkiya
Summary: Epilepsy is a common chronic neurological disorder, and using EEG signals for processing can improve the accuracy of seizure detection in neonates. The proposed CNN model showed high accuracy in predicting epileptic seizures in neonates, outperforming existing models.
JOURNAL OF INTELLIGENT & FUZZY SYSTEMS
(2021)
Article
Computer Science, Information Systems
R. Elakkiya, Kuppa Sai Sri Teja, L. Jegatha Deborah, Carmen Bisogni, Carlo Medaglia
Summary: Cervical cancer can be cured when diagnosed early, but accurate detection of cervical cells remains a challenge. A deep learning technique using SOD-GAN and F-SAE is proposed to automatically detect and classify premalignant and malignant conditions in cervical cells without preliminary assistance. Extensive experimentation shows promising improvement in efficiency and reduction in time complexity.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Information Systems
R. Elakkiya, Pandi Vijayakumar, Marimuthu Karuppiah
Summary: This study introduces a new method for COVID-19 screening using artificial intelligence and machine learning, and the experimental results show 100% accuracy in different datasets.
INFORMATION SYSTEMS FRONTIERS
(2021)
Article
Computer Science, Artificial Intelligence
R. Elakkiya, Pandi Vijayakumar, Neeraj Kumar
Summary: This paper introduces a method to classify sign language gestures using Generative Adversarial Networks, incorporating techniques such as stacked variational auto-encoders, Deep LSTM, and Deep Reinforcement Learning for hyperparameter optimization and regularization. Experimental results show that the proposed H-GANs improve accuracy and recognition rate compared to state-of-the-art methods.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Artificial Intelligence
B. Natarajan, R. Elakkiya, Moturi Leela Prasad
Summary: The development of Neural Machine Translation (NMT) systems has achieved significant progress in language translation tasks. This paper proposes a novel deep stacked GRU algorithm based NMT system to address the challenges in translating new words and out-of-vocabularies, and efficiently handle multilingual sentence translation tasks. The proposed model shows better translation results compared to earlier approaches, demonstrated through qualitative and quantitative evaluations.
JOURNAL OF AMBIENT INTELLIGENCE AND HUMANIZED COMPUTING
(2022)
Article
Public, Environmental & Occupational Health
R. Elakkiya, Deepak Kumar Jain, Ketan Kotecha, Sharnil Pandya, Sai Siddhartha Reddy, E. Rajalakshmi, Vijayakumar Varadarajan, Aniket Mahanti, Subramaniyaswamy Subramaniyaswamy
Summary: The field of bioinformatics has been rapidly developing over the past decade, with researchers using large amounts of data to extract biological knowledge. However, this has also led to issues with unbalanced data, such as how to classify precursor microRNA in RNA genome data. Experimental results show that the proposed Hybrid Deep Neural Network framework performs well in various genomes.
FRONTIERS IN PUBLIC HEALTH
(2021)
Article
Computer Science, Information Systems
R. Elakkiya, V. Subramaniyaswamy, V. Vijayakumar, Aniket Mahanti
Summary: The paper introduces an automated method for cervical cancer screening and diagnosis using digital colposcopy images, with the experimental results showing an accuracy of 99%.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2022)
Article
Computer Science, Artificial Intelligence
E. Rajalakshmi, R. Elakkiya, Alexey L. Prikhodko, M. G. Grif, Maxim A. Bakaev, Jatinderkumar R. Saini, Ketan Kotecha, V Subramaniyaswamy
Summary: The article discusses a Sign Language Recognition system for the hearing and vocally impaired population. A hybrid neural network architecture is proposed to address the challenges in recognizing isolated sign language from static and dynamic gestures. A novel dataset is created and experimental results show high accuracy in static and dynamic isolated sign recognition.
ACM TRANSACTIONS ON ASIAN AND LOW-RESOURCE LANGUAGE INFORMATION PROCESSING
(2023)
Article
Computer Science, Artificial Intelligence
B. Natarajan, R. Elakkiya
Summary: This paper introduces a novel generative framework called dynamic generative adversarial networks (dynamic GAN) for generating photo-realistic high-quality sign language videos. The model utilizes skeletal poses information and person images as input to produce high-quality videos through target frame generation and classification. Experimental evaluations show that the proposed model outperforms existing approaches.
Review
Energy & Fuels
Keerthana Sivamayil, Elakkiya Rajasekar, Belqasem Aljafari, Srete Nikolovski, Subramaniyaswamy Vairavasundaram, Indragandhi Vairavasundaram
Summary: This review paper analyzes 127 publications discussing the applications of Reinforcement Learning (RL) in various fields such as marketing, robotics, gaming, automated cars, natural language processing, internet of things security, recommendation systems, finance, and energy management. The focus is mainly on the RL application for energy management, which has proven beneficial in optimizing energy use in smart buildings, hybrid automobiles, smart grids, and managing renewable energy resources. RL is utilized to learn optimal control policies and make judgments based on sensor data, leading to reduced energy consumption and a sustainable environment. RL is also widely used in robotics, automated cars, gaming, security-related applications, and recommender systems. The article serves as a helpful resource for beginners to understand the foundations and applications of RL.
Article
Computer Science, Information Systems
E. Rajalakshmi, R. Elakkiya, V. Subramaniyaswamy, L. Prikhodko Alexey, Grif Mikhail, Maxim Bakaev, Ketan Kotecha, Lubna Abdelkareim Gabralla, Ajith Abraham
Summary: A novel vison-based hybrid deep neural net methodology is proposed in this study for recognizing Indian and Russian sign gestures. The proposed framework aims to establish a single framework for tracking and extracting multi-semantic properties, such as non-manual components and manual co-articulations. By using a 3D deep neural net with atrous convolutions for spatial feature extraction, attention-based Bi-LSTM for temporal and sequential feature extraction, modified autoencoders for abstract feature extraction, and a hybrid attention module for discriminative feature extraction, the proposed sign language recognition framework yields better results than other state-of-the-art frameworks.
Proceedings Paper
Computer Science, Cybernetics
Suriya T. Praba, S. Reka, R. Elakkiya
Summary: Poly Cystic Ovary Syndrome (PCOS) is a common hormonal disorder that affects a large number of women in their reproductive age and can lead to serious health issues. This research proposes an automated model using machine learning algorithms to classify PCOS and non-PCOS women, achieving 100% accuracy in predicting PCOS using Raman spectroscopy with follicular fluid samples.
2022 IEEE INTERNATIONAL SYMPOSIUM ON MIXED AND AUGMENTED REALITY ADJUNCT (ISMAR-ADJUNCT 2022)
(2022)
Article
Computer Science, Information Systems
B. Natarajan, E. Rajalakshmi, R. Elakkiya, Ketan Kotecha, Ajith Abraham, Lubna Abdelkareim Gabralla, V Subramaniyaswamy
Summary: This research proposes a novel approach for handling real-time sign language recognition, translation, and generation tasks, achieving high accuracy and visual quality using a hybrid model and optimized algorithms.