Search

Researchers at Taibah University in Saudi Arabia have developed a real-time recognition system for the American Sign Language alphabet using computer vision techniques. The system, which combines hand movement tracking with deep learning models, achieved high performance metrics with a precision of 98% and an F1 score of 99%. This study highlights the importance of sign language in communication for individuals who are deaf or hard of hearing, showcasing the potential for technology to enhance manual communication and rehabilitation therapy. [Extracted from the article]

The article discusses recent advances in machine learning for the processing of electromyographic signals in Polish Sign Language expressions. Researchers at Silesian University of Technology utilized convolutional neural networks to interpret sign language gestures based on EMG signals, achieving high accuracy rates. The study highlights the potential of integrating affordable EMG-based technologies for gesture recognition, offering a cost-effective solution for real-world applications and providing a basis for future research in EMG-based recognition of Polish Sign Language. [Extracted from the article]

Researchers from the National University La Plata have developed a low-parameter approach for Sign Language Recognition (SLR) using a ConvAtt Network model. By combining 1D convolutions with transformer layers, the model achieved a top-1 accuracy of 42.7% and a top-10 accuracy of 81.9% in categorizing signs of the French Belgian Sign Language. This approach is competitive with current state-of-the-art models while using significantly fewer parameters, making it a promising development in the field of computer science and therapy for the hard of hearing. [Extracted from the article]

A study conducted at the National University of La Plata (UNLP) focused on developing pose-based deep learning models for gloss-free Sign Language Translation (SLT). The research highlighted the challenges of using gloss annotations in SLT models due to limited data availability and the need for large models. The study proposed a gloss-free pose-based SLT model to reduce data dimensionality and model size, aiming to improve sign language translation accuracy. The research compared different models and developed a keypoint-based Transformer model for gloss-free SLT, showcasing promising results for future sign language translation advancements. [Extracted from the article]

Researchers from Shanghai University have developed a Structure-Aware Graph Convolutional Neural Network (SA-GNN) model for Continuous Sign Language Recognition (CSLR), aiming to improve social participation for deaf individuals. The model captures spatial structure and temporal variation of sign language motion units, enhancing weak supervision with a confidence cross-entropy loss. The study, supported by various foundations, achieved competitive results on popular CSLR datasets, contributing to the field of Information Processing and Management. [Extracted from the article]

A patent application by inventors Bean and Tokubo explores the translation of sign language in a virtual environment, focusing on methods, systems, and devices for this purpose. The application describes operations such as receiving video of a user, identifying signed communications in one sign language, translating them into another sign language, and rendering an avatar to display the translated communications. This innovation aims to facilitate real-time communication between users, particularly in the context of video games, offering a novel approach to enhancing accessibility and interaction for individuals using sign language. [Extracted from the article]

A patent application by Vittorio GIOVARA, filed on April 5, 2024, and made available online on October 10, 2024, describes a method for providing a sign-language avatar video for a primary video. The invention aims to improve accessibility to video display and streaming for those who communicate using sign language. The method involves displaying a 3D or 2D avatar alongside a video player window to replicate translated text or audio into sign language, using AI models to refine movements. The patent application, assigned to Vimeo.com Inc., highlights the use of accelerated graphical APIs and GPU processing calls for generating avatar animations. [Extracted from the article]

A patent application by inventors from California proposes a method for gesture translation in video games to facilitate communication between players. The technology involves analyzing player gestures, translating them into text based on the game context, and presenting the text communication in-game for other players to view. This innovation aims to enhance communication in multiplayer games, particularly for players using sign language or facing challenges in fast-paced gameplay. [Extracted from the article]

A study conducted by Chungbuk National University in South Korea focuses on improving sign language recognition in mobile environments to enhance communication between deaf and hearing individuals. The research proposes a technique to adjust the sampling rate dynamically based on the number of frames extracted during sign language utterances, aiming to accurately detect the start and end points of the signs. Results show that this dynamic sampling rate adjustment method significantly improves recognition performance compared to fixed sampling rate methods, highlighting its effectiveness in enhancing the accuracy and robustness of sign language recognition systems. The study underscores the importance of computational performance in mobile devices for effective sign language communication and rehabilitation of hearing-impaired individuals. [Extracted from the article]

A study conducted at Pir Mehr Ali Shah Arid Agriculture University in Rawalpindi, Pakistan, focused on developing a human hand gesture identification framework using SIFT and knowledge-level techniques. The research aimed to improve gesture recognition performance under varying lighting conditions by utilizing depth images collected from Microsoft Kinect devices. The study successfully achieved a 97.6% recognition accuracy rate for a dataset of 26 alphabets using an artificial neural network, demonstrating the effectiveness of the proposed method. This research contributes to the fields of engineering, machine learning, and rehabilitation, particularly in the context of sign language communication and therapy for hearing-impaired individuals. [Extracted from the article]

A study conducted by Beijing Normal University explored the impact of timing on students' learning achievement, satisfaction, and acceptance when using an augmented reality (AR) game in education. The research found that providing the AR game in class significantly improved learners' achievement and acceptance compared to providing it before class. While overall satisfaction was unaffected by timing, students' cognitive focus differed, with in-class participants emphasizing the game's educational value. Teachers and parents acknowledged the educational value of AR games but expressed concerns about costs and potential health issues, highlighting the need for careful consideration when implementing AR games in education. [Extracted from the article]

Researchers at the National Institute of Technology Calicut in Kerala, India, have developed a dynamic Indian Sign Language recognition system using contour-based hand tracking and maximum mean probability ensembling. This system aims to improve communication for individuals who rely on sign language by accurately recognizing double-handed dynamic gestures with an 89.83% accuracy rate. The study highlights the importance of manual communication and rehabilitation for the hard of hearing, emphasizing the need for accessible and cost-effective technologies in this field. The research, which has been peer-reviewed, showcases the potential of machine learning and emerging technologies in enhancing therapy and rehabilitation practices. [Extracted from the article]

A study conducted by East West University in Dhaka, Bangladesh, has developed an AI-based sensory glove system to recognize Bengali Sign Language (BaSL) for the speech-disabled community. The system incorporates sensors like flex, accelerometer, and gyroscopes to capture gestures and achieved an average recognition accuracy of 94.73% with a combination of CNN and LSTM networks. The research highlights the competitive features of the system in terms of latency, user comfort, and battery backup compared to other devices, offering potential benefits for the rehabilitation and communication of the hearing-impaired community. [Extracted from the article]

A recent study conducted at Tulane University explores the benefits and applications of video-span selection and search for real-time support in sign language video comprehension among American Sign Language (ASL) learners. The research aims to improve the video-watching experience for ASL learners by allowing them to select and view dictionary results for unfamiliar signs while watching a video. The study found that the integrated tool outperformed existing online dictionaries in terms of speed, accuracy, and workload, offering advantages for video comprehension. This research has implications for designers, computer vision researchers, and sign language educators. [Extracted from the article]

A recent systematic review conducted by researchers at Iran University of Medical Sciences explores the use of augmented reality (AR) and virtual reality (VR) in enhancing communication for individuals who are hard of hearing. The review found that these technologies have the potential to improve educational impacts, speech perception, and cognitive skills development among the deaf or hard of hearing population. The researchers emphasize the need for further research with larger sample sizes to fully investigate the benefits of AR and VR in supporting communication for individuals who are hard of hearing. [Extracted from the article]

Researchers at Beihang University in Beijing, China, have developed a model called MSF-ET (Multi-Scaled feature Fusion-Euclidean Transformer) to enhance continuous sign language recognition. The model integrates visual features of different granularities and refines the attention mechanism to improve understanding of sign language videos. By addressing limitations in existing models, the researchers aim to enhance dynamic temporal modeling capabilities for accurate sign language recognition. This study contributes to the field of applied sciences and has implications for therapy, manual communication, and the rehabilitation of hearing-impaired individuals. [Extracted from the article]

A recent study from Zagazig University in Egypt focused on developing a novel approach to recognizing Arabic sign alphabets using optimized hybrid techniques. The study combined a convolutional neural network with traditional machine learning algorithms and achieved high classification accuracy of nearly 99% on a dataset of 32 different classes of Arabic alphabet letters. The research aims to address communication barriers faced by hard of hearing individuals, highlighting the importance of bridging communication gaps for improved education and employment opportunities. The study has been peer-reviewed and offers valuable insights into the potential of machine learning in sign language recognition. [Extracted from the article]

Researchers from City College of New York are exploring the integration of Artificial Intelligence (AI) with sign language recognition to address communication barriers for the deaf and hard-of-hearing communities. The study discusses the evolution of sign language recognition technology and emphasizes the importance of government and industry support, ethical data practices, and user-centered design in advancing this inclusive technology. The research highlights the benefits of multi-modal AI models in enhancing recognition accuracy and underscores the need for user-focused solutions and innovative technical approaches in integrating this technology into daily life. [Extracted from the article]

A recent report discusses new research findings in applied sciences related to sign language interpreting systems. The research focuses on developing a system based on Recursive Neural Networks (RNN) to predict the words that a person intends to communicate in Mexican Sign Language (MSL). The researchers trained four RNN-based models using a dataset of 600 clips and conducted experiments to measure the system's accuracy in offline and online modes. The results indicate a high accuracy of 0.93 in offline mode and a better performance compared to previous approaches in online mode. The proposed system has potential applications in teaching, learning, commercial transactions, and daily communications between deaf and non-deaf individuals. [Extracted from the article]

A recent study conducted by researchers at the Chinese University of Hong Kong compared the count-mass distinction (CMD) in Hong Kong Sign Language (HKSL) to American Sign Language (ASL). The study found that the existing diagnostics used in spoken language research are insufficient to identify CMDs in sign languages. HKSL and ASL differ in several ways, including their diagnostic criteria for CMD. HKSL is classified as a number marking language (Type I) and exhibits a failure to combine mass nouns directly with numerals and count adjectives without the intervention of a classifier. The study concluded that HKSL is a number marking language with plural number marking realized through zero marking. This research has been peer-reviewed and provides valuable insights into the linguistic characteristics of HKSL. [Extracted from the article]

A recent report from the School of Computer Science and Engineering discusses the advancements in deep neural networks for lipreading with sign language recognition. Lip reading, also known as speech reading, is a visual form of communication that involves observing the speaker's face and interpreting their lip movements. The report explores the mechanisms and techniques involved in lip reading, including face identification, lip localization, feature extraction, and identifying phrases or sentences based on lip movements. It also discusses the integration of sign language with lip reading and highlights the challenges and future research directions in this field. For more information, the full article can be accessed through IEEE Access. [Extracted from the article]

A recent study conducted at the University of Chicago explored how certainty is expressed in Turkish Sign Language (TID). The researchers found that signers in TID use both manual (hands) signs and nonmanuals (face and body movements) to convey their certainty. The study also revealed that certain nonmanual cues, such as head nods and squints, can affect the threshold of certainty. This research provides further insight into the dynamic nature of certainty expression in sign languages. For more information, readers can refer to the article "Simultaneity of Certainty In Turkish Sign Language (Tid)" published in the Journal of Pragmatics. [Extracted from the article]

A recent study published by the Department of Electronics and Automation introduces a new system for recognizing and regenerating hand gestures using a wearable glove equipped with flex sensors and a MEMS-based accelerometer array. The system utilizes a multilayer perceptron feed-forward neural network algorithm to determine the gestures, and the accuracy of the recognition was found to be 99.67% based on measured test data. The research suggests that this method improves sign language recognition and enhances the tracking of hand movements compared to existing methods. [Extracted from the article]

Researchers from Shamoon College of Engineering in Be'er Sheva, Israel, have developed an advanced Israeli Sign Language (ISL) recognition system that can interpret dynamic motion gestures in real-time video. Unlike conventional systems that focus on static signs, this system incorporates deep learning and Computer Vision techniques to bridge communication gaps between the deaf community and the broader population. The researchers provide a comprehensive account of their preprocessing pipeline and dataset, which is publicly accessible, allowing for future research and the deployment of the model on a publicly accessible website. The system aims to promote inclusivity and understanding. [Extracted from the article]

Researchers from King Abdulaziz University in Saudi Arabia have developed a multimodal dataset for Arabic Sign Language (ArSL) production. The dataset includes audio signals, texts, and videos recorded by expert signers and annotated using gloss representation. The researchers used various evaluation methods to assess the accuracy and understanding of the ArSL videos. The study highlights the importance of careful annotation and representation in ArSL recognition systems and aims to bridge the gap in accessible religious content for the deaf community. [Extracted from the article]

A recent study conducted at Kafrelsheikh University in Egypt has developed a real-time Arabic avatar system to assist deaf and nonspeaking deaf individuals in communicating with others. The system uses deep learning sign language translation to translate text or spoken input into Arabic Sign Language (ArSL) movements that the avatar makes. The avatar is trained on diverse datasets of Arabic sign language images, allowing it to accurately capture the nuances of hand movements. The study found that the suggested approach improved communication effectiveness and ease for deaf and nonspeaking deaf individuals in Arabic-speaking communities. This research has been peer-reviewed and provides valuable insights into improving communication for hard of hearing people. [Extracted from the article]

A recent study from King Abdulaziz University in Jeddah, Saudi Arabia, explores advancements in sign language recognition. The study highlights the importance of sign language as a mode of communication for the hard of hearing and the need for innovative solutions to bridge the gap between hearing and non-hearing communities. The researchers found that deep learning techniques, such as convolutional and recurrent neural networks, have shown high accuracy in sign language recognition, and integrating non-manual features has improved recognition accuracy. The study concludes that ongoing advancements in this field have the potential to revolutionize communication and break down barriers for individuals who rely on sign language. [Extracted from the article]

A recent study conducted at Debre Markos University in Ethiopia focused on the recognition of derived Amharic alphabet sign language using machine learning techniques. The researchers aimed to bridge the communication gap between hearing-impaired individuals and those without hearing disabilities. They proposed the use of Support Vector Machine (SVM) with Convolutional Neural Network (CNN) and Histogram of Oriented Gradients (HOG) for feature extraction and classification. The results showed that the SVM model achieved an accuracy of 89.02% to 97.40% using different hybrid feature vectors, while the CNN (softmax) classifier achieved a 93.55% accuracy. This research contributes to the development of technologies that can facilitate communication with hearing-impaired individuals. [Extracted from the article]

A researcher at Gallaudet University has developed an innovative informed consent app for deaf, deafblind, and hard of hearing individuals who use sign language and wish to participate in cancer clinical trials. The app displays each section of the informed consent form in sign language and captures the participant's signed responses through a mobile device camera. Participants can respond with "yes," "no," or "repeat" if they did not understand a specific section. The app also allows participants to record their signature in sign language. Feedback from diverse signers showed that the app is culturally and linguistically appropriate for people with sensory disabilities. This technology aims to increase the inclusion of sign language users in clinical trials and give them greater autonomy as research participants. The study was funded by the National Library of Medicine/NIH. [Extracted from the article]

A patent application titled "Systems And Methods For Facilitating Private Communication Between Users" has been published online. The invention aims to enable private communication between users during the execution of a computer program, such as a video game or a sign language program. The system involves capturing real-world gestures, such as sign communications, and associating them with virtual gestures that are only visible to the intended recipient. This technology allows users to communicate strategy and meanings without other players or users being aware of the communication. The patent application was filed by Celeste Bean from San Francisco, CA. [Extracted from the article]

Samsung Electronics Co. Ltd. has been issued a patent for an electronic device that provides subtitles and sign language for deaf individuals by automatically recognizing speech in content. The device includes a display, memory, and processor that can obtain an expected word from the content, perform speech recognition using the expected word, and display the converted text. The device can also provide sign language images corresponding to the text. This invention aims to increase accuracy in speech recognition without relying on a large database and improve accessibility for the hearing impaired. [Extracted from the article]

A patent has been issued to T-Mobile USA Inc. for a system that aims to improve communication accessibility for individuals with hearing disabilities. The system addresses the issue of using separate numbers for voice and other forms of communication, such as text messaging, by implementing a single number for all types of communication. This system involves the use of a video relay service (VRS) or video interpreting service (VIS) to facilitate voice calls between hearing users and individuals with hearing disabilities. The patent describes the technical aspects of the system, including the use of session border controllers (SBCs) and assistance routing data. [Extracted from the article]

Researchers from the Department of Electrical and Communication Engineering in Tamil Nadu, India, have developed a deep learning-based hand gesture recognition model to address communication difficulties faced by deaf-dumb individuals. The model utilizes the Hybrid Convolution-based EfficientNet (HCEN) framework, which extracts features from hand gesture images using the Scale-Invariant Feature Transform (SIFT) method and Multiscale Vision Transformer (MVT). The HCEN model is optimized using the Adaptive Political Optimizer (APO) algorithm. The researchers conducted various experiments to evaluate the performance of the proposed model. This research has been peer-reviewed and published in the International Journal of Pattern Recognition and Artificial Intelligence. [Extracted from the article]

A recent study conducted at Ohio State University examined the development of English reading skills in deaf and hard of hearing (d/Dhh) children and adolescents. The study compared two frameworks, the Qualitative Similarity Hypothesis (QSH) and the American Sign Language (ASL)-English sign-print bilingual model, to determine the most effective approach. The research found that while there are benefits to using American Sign Language in a sign-print ASL-English bilingual education program, these benefits diminish during the later early literacy period. The study suggests that the QSH and other developmental models provide stronger evidence and a better explanation for the difficulties faced by d/Dhh students in developing English reading skills. Further research is recommended. [Extracted from the article]

A report from the United Arab Emirates University discusses the need for an efficient sign language translation system due to the shortage of certified sign language interpreters. The research provides an analysis of the evolution of sign language machine translation algorithms and a taxonomy of the most commonly used approach, Transformers architectures. The study also presents the requirements for a real-time sign language machine translation system supported by accurate deep learning algorithms. The research proposes future directions for sign language translation systems. [Extracted from the article]

A study conducted at the University of Colorado Boulder examined the long-term neurocognitive outcomes in young adults with cochlear implants (CI). The study compared the cognitive outcomes of early-implanted and late-implanted CI users, as well as typically hearing controls. The results showed no significant differences in cognition and nonverbal intelligence between the early and late-implanted groups, but the late-implanted group had higher American Sign Language (ASL) comprehension. The study suggests that early access to language, whether spoken or signed, is important for cognitive development, and crossmodal plasticity does not necessarily impact cognitive outcomes. [Extracted from the article]

A recent study conducted by researchers from the University of Ljubljana in Slovenia focused on the inclusion of the Slovenian Sign Language (SZJ) community in language research, learning, and teaching. The study found that deaf users of Slovenian Sign Language vary in their competence levels, with some having delayed or limited exposure to the language. The researchers aimed to address this by creating a reference grammar of SZJ and training deaf signers as researchers and teachers. The study concluded that involving signers in the creation and dissemination of the reference grammar was successful, as evidenced by the continuous viewership of the materials by both deaf and hearing users. [Extracted from the article]

A patent application for a portable assistive device for challenged individuals has been filed by inventors Aishwarya Karnataki and Parikshit Sohoni. The device aims to address the communication difficulties faced by deaf individuals who rely on sign language. Existing devices that convert hand gestures to text or speech are limited in their capabilities, and there is a need for a device that enables two-way communication between differently abled individuals and normal people. The portable assistive device includes wearable devices, a handheld device, and a server that utilize machine learning and artificial intelligence algorithms to process gesture data and convert it into relevant text or speech. [Extracted from the article]

A recent study conducted by researchers from CIESAS in Mexico City explores the cognitive representation of space in the Yucatec Maya Sign Language (YSML). The YSML is an emerging visual language created by deaf and hearing signers in Yucatec Maya villages. The study found that YSML signers share a preference for the geocentric Frame of Reference (FoR), which is consistent with the preference observed in spoken Yucatec Maya. The research suggests that cultures around the world rely on a preferred spatial FoR, which influences how people orient themselves and think about the world. [Extracted from the article]

A recent report from Stockholm University discusses the challenges faced by adult deaf migrants in Sweden when it comes to language learning. Upon arrival in Sweden, all adult migrants, including those who are deaf, are required to learn Swedish. However, deaf migrants face unique challenges due to a lack of interpreters proficient in different sign languages and limited exposure to language and education. The report emphasizes the importance of addressing the specific language-learning needs of deaf migrants to facilitate their successful integration into Swedish society. This research was supported by the Swedish Research Council and has been peer-reviewed. [Extracted from the article]

Researchers from Tianjin University in China have developed a Multi-Scale Context-Aware network (MSCA-Net) to improve continuous sign language recognition (CSLR) methods. The MSCA-Net consists of two main modules: Multi-Scale Motion Attention (MSMA) and Multi-Scale Temporal Modeling (MSTM). The MSMA module uses the differences among frames to perceive information of the hands and face in multiple spatial scales, while the MSTM module explores crucial temporal information in sign language videos from different temporal scales. The researchers conducted extensive experiments using three sign language datasets and found that the MSCA-Net achieved state-of-the-art performance. [Extracted from the article]