The predominant means of communication is speech; however, there are persons whose speaking or hearing abilities are impaired. Communication presents a significant barrier for persons with such disabilities. The use of deep learning methods can help to reduce communication barriers. This paper proposes a deep learning-based model that detects and recognizes the words from a person’s gestures. Deep learning models, namely, LSTM and GRU (feedback-based learning models), are used to recognize signs from isolated Indian Sign Language (ISL) video frames. The four different sequential combinations of LSTM and GRU (as there are two layers of LSTM and two layers of GRU) were used with our own dataset, IISL2020. The proposed model, consisting of a single layer of LSTM followed by GRU, achieves around 97% accuracy over 11 different signs. This method may help persons who are unaware of sign language to communicate with persons whose speech or hearing is impaired.
Sign language is the most common form of communication for the deaf and dumb. To bridge the communication gap with such impaired people, normal people should be able to recognize signs. Therefore, it is necessary to introduce a sign language recognition system to assist such impaired people. This paper proposes the Transformer Encoder as a useful tool for sign language recognition. For the recognition of static Indian signs, the authors have implemented a vision transformer. To recognize static Indian sign language, proposed methodology archives noticeable performance over other state-of-the-art convolution architecture. The suggested methodology divides the sign into a series of positional embedding patches, which are then sent to a transformer block with four self-attention layers and a multilayer perceptron network. Experimental results show satisfactory identification of gestures under various augmentation methods. Moreover, the proposed approach only requires a very small number of training epochs to achieve 99.29 percent accuracy.
Liveness detection for fingerprint impressions plays a role in the meaningful prevention of any unauthorized activity or phishing attempt. The accessibility of unique individual identification has increased the popularity of biometrics. Deep learning with computer vision has proven remarkable results in image classification, detection, and many others. The proposed methodology relies on an attention model and ResNet convolutions. Spatial attention (SA) and channel attention (CA) models were used sequentially to enhance feature learning. A three-fold sequential attention model is used along with five convolution learning layers. The method’s performances have been tested across different pooling strategies, such as Max, Average, and Stochastic, over the LivDet-2021 dataset. Comparisons against different state-of-the-art variants of Convolutional Neural Networks, such as DenseNet121, VGG19, InceptionV3, and conventional ResNet50, have been carried out. In particular, tests have been aimed at assessing ResNet34 and ResNet50 models on feature extraction by further enhancing the sequential attention model. A Multilayer Perceptron (MLP) classifier used alongside a fully connected layer returns the ultimate prediction of the entire stack. Finally, the proposed method is also evaluated on feature extraction with and without attention models for ResNet and considering different pooling strategies.
The issue of security is paramount in any organisation. Therefore, the authors intend to aid in the security of such organisations by bringing a video based human authentication system for access control which is a type of cyber physical system (CPS). CPS is an integration of computation and physical processes; here the computation is provided by face detection and recognition algorithm and physical process is the input human face. This system aims to provide a platform that allows any authorized person to enter or leave the premise automatically by using face detection and recognition technology. The system also provides the administrator with the access to the logs, wherein he/she would be able to access the details of the people entering or leaving the organisation along with the live video streaming so that there is no sneaking of any unauthorized person with any other authorized person. The administrator can also do registration on behalf of a new person who requires access to the premises for a restricted amount of time only as specified by the administrator.
A medical disorder known as diabetic retinopathy (DR) affects people who suffer from diabetes. Globally many people visually impaired and blind due to diabetic retinopathy. The primary cause of DR in diabetic patients is high blood sugar and it affects blood vessels available in the retinal cell. The recent advancement in deep learning and computer vision methods, and its automation applications able to recognize present of DR in retinal cells and vessel images. Authors have proposed Attention-based hybrid model to extract features. Proposed methodology uses DenseNet121 architecture for convolution learning and then the feature vector will be enhanced with channel and spatial attention model. The proposed architecture also simulates for binary and multiclass classification to recognized infection and spreading of disease. Binary classification recognize DR images either positive or negative, while multiclass classification represents an infection in a scale of 0 to 4. Simulation of the proposed methodology has achieved 98.57 % and 99.01% accuracy for multiclass and binary classification, respectively. Simulation of the study also explored the impact of data augmentation to make the proposed model robust and generalized. Attention based deep learning model has achieve remarkable accuracy to detect diabetic infection from retinal cellular images.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.