Search

Human activity recognition (HAR) for disabled people is a vital research area, which aims to help individuals with disabilities in their daily lives. HAR involves using technology, typically wearable devices or sensors, to automatically identify and classify human activities and movements. HAR using deep learning (DL) is an effective and popular method to automatically classify and identify human activities based on sensor information. This article develops a hyperparameter tuned fuzzy deep neural network–based HAR (HTFDNN‐HAR) method. The objective of the HTFDNN‐HAR method lies in human activities identification and classification. In the presented HTFDNN‐HAR technique, cross‐guided bilateral filtering (CGBF)–based preprocessing is initially applied and MobileNetV3 architecture is applied for the effectual extraction of the feature vectors. In addition, the HTFDNN‐HAR technique makes use of the FDNN method for the efficient detection and classification of human activities. Finally, the HTFDNN‐HAR technique applies the manta ray foraging optimization (MRFO) technique for the optimum hyperparameter selection of the FDNN approach. A wide range of experiments have been carried out on the CAUCAFall dataset comprising 10,000 samples with two classes. The simulation value highlighted that the HTFDNN‐HAR technique reaches an enhanced accuracy of 99.40% over other recent approaches. [ABSTRACT FROM AUTHOR]

Fall detection (FD) for disabled persons in the Internet of Things (IoT) platform contains a combination of sensor technologies and data analytics for automatically identifying and responding to samples of falls. In this regard, IoT devices like wearable sensors or ambient sensors from the personal space role a vital play in always monitoring the user's movements. FD employs deep learning (DL) in an IoT platform using sensors, namely accelerometers or depth cameras, to capture data connected to human movements. DL approaches are frequently recurrent neural networks (RNNs) or convolutional neural networks (CNNs) that have been trained on various databases for recognizing patterns connected with falls. The trained methods are then executed on edge devices or cloud environments for real-time investigation of incoming sensor data. This method differentiates normal activities and potential falls, triggering alerts and reports to caregivers or emergency numbers once a fall is identified. We designed an Artificial Rabbit Optimizer with a DL-based FD and classification (ARODL-FDC) system from the IoT environment. The ARODL-FDC approach proposes to detect and categorize fall events to assist elderly people and disabled people. The ARODL-FDC technique comprises a fourstage process. Initially, the preprocessing of input data is performed by Gaussian filtering (GF). The ARODL-FDC technique applies the residual network (ResNet) model for feature extraction purposes. Besides, the ARO algorithm has been utilized for better hyperparameter choice of the ResNet algorithm. At the final stage, the full Elman Neural Network (FENN) model has been utilized for the classification and recognition of fall events. The experimental results of the ARODL-FDC technique can be tested on the fall dataset. The simulation results inferred that the ARODL-FDC technique reaches promising performance over compared models concerning various measures. [ABSTRACT FROM AUTHOR]

Text-To-Speech (TTS) is a speech processing tool that is highly helpful for visually-challenged people. The TTS tool is applied to transform the texts into human-like sounds. However, it is highly challenging to accomplish the TTS outcomes for the non-diacritized text of the Arabic language since it has multiple unique features and rules. Some special characters like gemination and diacritic signs that correspondingly indicate consonant doubling and short vowels greatly impact the precise pronunciation of the Arabic language. But, such signs are not frequently used in the texts written in the Arabic language since its speakers and readers can guess them from the context itself. In this background, the current research article introduces an Optimal Deep Learning-driven Arab Text-to-Speech Synthesizer (ODLD-ATSS) model to help the visually-challenged people in the Kingdom of Saudi Arabia. The prime aim of the presented ODLD-ATSS model is to convert the text into speech signals for visually-challenged people. To attain this, the presented ODLD-ATSS model initially designs a Gated Recurrent Unit (GRU)-based prediction model for diacritic and gemination signs. Besides, the Buckwalter code is utilized to capture, store and display the Arabic texts. To improve the TSS performance of the GRU method, the Aquila Optimization Algorithm (AOA) is used, which shows the novelty of the work. To illustrate the enhanced performance of the proposed ODLD-ATSS model, further experimental analyses were conducted. The proposed model achieved a maximum accuracy of 96.35%, and the experimental outcomes infer the improved performance of the proposed ODLD-ATSS model over other DL-based TSS models. [ABSTRACT FROM AUTHOR]

Mobile communication and the Internet of Things (IoT) technologies have recently been established to collect data from human beings and the environment. The data collected can be leveraged to provide intelligent services through different applications. It is an extreme challenge to monitor disabled people from remote locations. It is because day-to-day events like falls heavily result in accidents. For a person with disabilities, a fall event is an important cause of mortality and post-traumatic complications. Therefore, detecting the fall events of disabled persons in smart homes at early stages is essential to provide the necessary support and increase their survival rate. The current study introduces a Whale Optimization Algorithm Deep Transfer Learning-Driven Automated Fall Detection (WOADTL-AFD) technique to improve the Quality of Life for persons with disabilities. The primary aim of the presented WOADTL-AFD technique is to identify and classify the fall events to help disabled individuals. To attain this, the proposed WOADTL-AFD model initially uses a modified SqueezeNet feature extractor which proficiently extracts the feature vectors. In addition, the WOADTLAFD technique classifies the fall events using an extreme Gradient Boosting (XGBoost) classifier. In the presented WOADTL-AFD technique, the WOA approach is used to fine-tune the hyperparameters involved in the modified SqueezeNet model. The proposed WOADTL-AFD technique was experimentally validated using the benchmark datasets, and the results confirmed the superior performance of the proposed WOADTL-AFD method compared to other recent approaches. [ABSTRACT FROM AUTHOR]