Search Results (30)

Following the COVID-19 pandemic, public interest in healthcare has significantly in-creased, emphasizing the importance of early disease detection through lung sound analysis. Lung sounds serve as a critical biomarker in the diagnosis of pulmonary diseases, and numerous deep learning-based approaches have been actively explored for this purpose. Existing lung sound classification models have demonstrated high accuracy, benefiting from recent advances in artificial intelligence (AI) technologies. However, these models often rely on transmitting data to computationally intensive servers for processing, introducing potential security risks due to the transfer of sensitive medical information over networks. To mitigate these concerns, on-device AI has garnered growing attention as a promising solution for protecting healthcare data. On-device AI enables local data processing and inference directly on the device, thereby enhancing data security compared to server-based schemes. Despite these advantages, on-device AI is inherently limited by computational constraints, while conventional models typically require substantial processing power to maintain high performance. In this study, we propose a lightweight lung sound classification model designed specifically for on-device environments. The proposed scheme extracts audio features using Mel spectrograms, chromagrams, and Mel-Frequency Cepstral Coefficients (MFCC), which are converted into image representations and stacked to form the model input. The lightweight model performs convolution operations tailored to both temporal and frequency–domain characteristics of lung sounds. Comparative experimental results demonstrate that the proposed model achieves superior inference performance while maintaining a significantly smaller model size than conventional classification schemes, making it well-suited for deployment on resource-constrained devices. Full article

Continuous-time Recurrent Neural Networks (CTRNNs) are well-suited for modeling temporal dynamics in low-power neuromorphic and analog computing systems, making them promising candidates for edge-based human activity recognition (HAR) in healthcare. However, training CTRNNs remains challenging due to their continuous-time nature and the need to respect physical hardware constraints. In this work, we propose G-CTRNN, a novel gradient-based training framework for analog-friendly CTRNNs designed for embedded healthcare applications. Our method extends Backpropagation Through Time (BPTT) to continuous domains using TensorFlow’s automatic differentiation, while enforcing constraints on time constants and synaptic weights to ensure hardware compatibility. We validate G-CTRNN on the WISDM human activity dataset, which simulates realistic wearable sensor data for healthcare monitoring. Compared to conventional RNNs, G-CTRNN achieves superior classification accuracy with fewer parameters and greater stability—enabling continuous, real-time HAR on low-power platforms such as MEMS computing networks. The proposed framework provides a pathway toward on-device AI for remote patient monitoring, elderly care, and personalized healthcare in resource-constrained environments. Full article

Salivary $\alpha$-amylase (sAA) is a widely recognized biomarker for stress and autonomic nervous system activity. However, conventional enzymatic assays used to quantify sAA are limited by time-consuming, lab-based protocols. In this study, we present a portable, AI-driven point-of-care system for automated sAA classification via colorimetric image analysis. The system integrates SCHEDA, a custom-designed imaging device providing and ensuring standardized illumination, with a deep learning pipeline optimized for mobile deployment. Two classification strategies were compared: (1) a modular YOLOv4-CNN architecture and (2) a unified YOLOv8 segmentation-classification model. The models were trained on a dataset of 1024 images representing an eight-class classification problem corresponding to distinct sAA concentrations. The results show that red-channel input significantly enhances YOLOv4-CNN performance, achieving 93.5% accuracy compared to 88% with full RGB images. The YOLOv8 model further outperformed both approaches, reaching 96.5% accuracy while simplifying the pipeline and enabling real-time, on-device inference. The system was deployed and validated on a smartphone, demonstrating consistent results in live tests. This work highlights a robust, low-cost platform capable of delivering fast, reliable, and scalable salivary diagnostics for mobile health applications. Full article

Communication barriers between deaf and hearing individuals present significant challenges to social inclusion, highlighting the need for effective technological aids. This study aimed to bridge this gap by developing a mobile system for the real-time translation of Portuguese Sign Language (LGP) alphabet gestures into text, addressing a specific technological void for LGP. The core of the solution is a mobile application integrating two distinct machine learning approaches trained on a custom LGP dataset: firstly, a Convolutional Neural Network (CNN) optimized with TensorFlow Lite for efficient, on-device image classification, enabling offline use; secondly, a method utilizing MediaPipe for hand landmark extraction from the camera feed, with classification performed by a server-side Multilayer Perceptron (MLP). Evaluation tests confirmed that both approaches could recognize LGP alphabet gestures with good accuracy (F1-scores of approximately 76% for the CNN and 77% for the MediaPipe+MLP) and processing speed (1 to 2 s per gesture on high-end devices using the CNN and 3 to 5 s under typical network conditions using MediaPipe+MLP), facilitating efficient real-time translation, though performance trade-offs regarding speed versus accuracy under different conditions were observed. The implementation of this dual-method system provides crucial flexibility, adapting to varying network conditions and device capabilities, and offers a scalable foundation for future expansion to include more complex gestures. This work delivers a practical tool that may contribute to improve communication accessibility and the societal integration of the deaf community in Portugal. Full article

The spread of biometric systems utilizing fingerprints has increased the need for advanced spoof detection techniques, but training convolutional neural networks (CNNs) with the limited number of images available in fingerprint datasets poses significant challenges. In this paper, we propose a lightweight network architecture which addresses the challenges inherent in small fingerprint datasets by employing a moderately deep network architecture which is sufficient for extracting essential features from fingerprint images. We apply a hyperbolic tangent activation to the final feature map, which has features from local receptive fields, and average the responses into a single value. Thus, our architecture reduces overfitting by increasing the number of effective labels during training. Additionally, the incorporation of the spatial attention module enhances feature representation, culminating in improved accuracy. The evaluation results show that the proposed model, with only 0.14 million parameters, outperforms existing techniques including lightweight models and transfer-learning-based models, achieving superior average test accuracies of 98.30% and 95.57% on the LivDet-2015 and -2017 datasets, respectively. It also delivers state-of-the-art cross-material performance, with corresponding average classification error values of 0.81% and 1.91%, making it highly effective for on-device fingerprint authentication. Full article

The deployment of machine learning models on mobile platforms has ushered in a new era of innovation across diverse sectors, including agriculture, where such applications hold immense promise for empowering farmers with cutting-edge technologies. In this context, the threat posed by insects to crop yields during harvest has escalated, fueled by factors such as evolution and climate change-induced shifts in insect behavior. To address this challenge, smart insect monitoring systems and detection models have emerged as crucial tools for farmers and IoT-based systems, enabling interventions to safeguard crops. The primary contribution of this study lies in its systematic investigation of model optimization techniques for edge deployment, including Post-Training Quantization, Quantization-Aware Training, and Data Representative Quantization. As such, we address the crucial need for efficient, on-site pest detection tools in agricultural settings. We provide a detailed analysis of the trade-offs between model size, inference speed, and accuracy across different optimization approaches, ensuring practical applicability in resource-constrained farming environments. Our study explores various methodologies for model development, including the utilization of Mobile-ViT and EfficientNet architectures, coupled with transfer learning and fine-tuning techniques. Using the Dangerous Farm Insects Dataset, we achieve an accuracy of 82.6% and 77.8% on validation and test datasets, respectively, showcasing the efficacy of our approach. Furthermore, we investigate quantization techniques to optimize model performance for on-device inference, ensuring seamless deployment on mobile devices and other edge devices without compromising accuracy. The best quantized model, produced through Post-Training Quantization, was able to maintain a classification accuracy of 77.8% while significantly reducing the model size from 33 MB to 9.6 MB. To validate the generalizability of our solution, we extended our experiments to the larger IP102 dataset. The quantized model produced using Post-Training Quantization was able to maintain a classification accuracy of 59.6% while also reducing the model size from 33 MB to 9.6 MB, thus demonstrating that our solution maintains a competitive performance across a broader range of insect classes. Full article

Skin cancer is one of the most common and life-threatening diseases. In the current era, early detection remains a significant challenge, particularly in remote and underserved regions with limited internet access. Traditional skin cancer detection systems often depend on image classification using deep learning models that require constant connectivity to internet access, creating barriers in areas with poor infrastructure. To address this limitation, CNN provides an innovative solution by enabling on-device machine learning on low-computing Internet of Things (IoT) devices. This study evaluates the performance of a convolutional neural network (CNN) model trained on 10,000 dermoscopic images spanning seven classes from the Harvard Skin Lesion dataset. Unlike previous research, which seldom offers detailed performance evaluations on IoT hardware, this work benchmarks the CNN model on multiple single-board computers (SBCs), including low-computing devices like Raspberry Pi and Jetson Nano. The evaluation focuses on classification accuracy and hardware efficiency, analyzing the impact of varying training dataset sizes to assess the model’s scalability and effectiveness on resource-constrained devices. The simulation results demonstrate the feasibility of deploying accurate and efficient skin cancer detection systems directly on low-power hardware. The simulation results show that our proposed method achieves an accuracy of 98.25%, with the fastest hardware being the Raspberry Pi 5, which achieves a detection time of 0.01 s. Full article

Pressure ulcers are serious healthcare concerns, especially for the elderly with reduced mobility. Severe pressure ulcers are accompanied by pain, degrading patients’ quality of life. Thus, speedy and accurate detection and classification of pressure ulcers are vital for timely treatment. The conventional visual examination method requires professional expertise for diagnosing pressure ulcer severity but it is difficult for the lay carer in domiciliary settings. In this study, we present a mobile healthcare platform incorporated with a light-weight deep learning model to exactly detect pressure ulcer regions and classify pressure ulcers into six severities such as stage 1–4, deep tissue pressure injury, and unstageable. YOLOv8 models were trained and tested using 2800 annotated pressure ulcer images. Among the five tested YOLOv8 models, the YOLOv8m model exhibited promising detection performance with overall classification accuracy of 84.6% and a mAP@50 value of 90.8%. The mobile application (app) was also developed applying the trained YOLOv8m model. The mobile app returned the diagnostic result within a short time (≒3 s). Accordingly, the proposed on-device AI app can contribute to early diagnosis and systematic management of pressure ulcers. Full article

This study proposes an edge computing-based facial expression recognition system that is low cost, low power, and privacy preserving. It utilizes a minimally obtrusive cap-based system designed for the continuous and real-time monitoring of a user’s facial expressions. The proposed method focuses on detecting facial skin deformations accompanying changes in facial expressions. A multi-zone time-of-flight (ToF) depth sensor VL53L5CX, featuring an 8 × 8 depth image, is integrated into the front brim of the cap to measure the distance between the sensor and the user’s facial skin surface. The distance values corresponding to seven universal facial expressions (neutral, happy, disgust, anger, surprise, fear, and sad) are transmitted to a low-power STM32F476 microcontroller (MCU) as an edge device for data preprocessing and facial expression classification tasks utilizing an on-device pre-trained deep learning model. Performance evaluation of the system is conducted through experiments utilizing data collected from 20 subjects. Four deep learning algorithms, including Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM) networks, and Deep Neural Networks (DNN), are assessed. These algorithms demonstrate high accuracy, with CNN yielding the best result, achieving an accuracy of 89.20% at a frame rate of 15 frames per second (fps) and a maximum latency of 2 ms. Full article

This paper presents an on-device semi-supervised human activity detection system that can learn and predict human activity patterns in real time. The clinical objective is to monitor and detect the unhealthy sedentary lifestyle of a user. The proposed semi-supervised learning (SSL) framework uses sparsely labelled user activity events acquired from Inertial Measurement Unit sensors installed as wearable devices. The proposed cluster-based learning model in this approach is trained with data from the same target user, thus preserving data privacy while providing personalized activity detection services. Two different cluster labelling strategies, namely, population-based and distance-based strategies, are employed to achieve the desired classification performance. The proposed system is shown to be highly accurate and computationally efficient for different algorithmic parameters, which is relevant in the context of limited computing resources on typical wearable devices. Extensive experimentation and simulation study have been conducted on multi-user human activity data from the public domain in order to analyze the trade-off between classification accuracy and computation complexity of the proposed learning paradigm with different algorithmic hyper-parameters. With 4.17 h of training time for 8000 activity episodes, the proposed SSL approach consumes at most 20 KB of CPU memory space, while providing a maximum accuracy of 90% and 100% classification rates. Full article

Electrocardiography (ECG) has emerged as a ubiquitous diagnostic tool for the identification and characterization of diverse cardiovascular pathologies. Wearable health monitoring devices, equipped with on-device biomedical artificial intelligence (AI) processors, have revolutionized the acquisition, analysis, and interpretation of ECG data. However, these systems necessitate AI processors that exhibit flexible configuration, facilitate portability, and demonstrate optimal performance in terms of power consumption and latency for the realization of various functionalities. To address these challenges, this study proposes an instruction-driven convolutional neural network (CNN) processor. This processor incorporates three key features: (1) An instruction-driven CNN processor to support versatile ECG-based application. (2) A Processing element (PE) array design that simultaneously considers parallelism and data reuse. (3) An activation unit based on the CORDIC algorithm, supporting both Tanh and Sigmoid computations. The design has been implemented using 110 nm CMOS process technology, occupying a die area of 1.35 mm² with 12.94 µW power consumption. It has been demonstrated with two typical ECG AI applications, including two-class (i.e., normal/abnormal) classification and five-class classification. The proposed 1-D CNN algorithm performs with a 97.95% accuracy for the two-class classification and 97.9% for the five-class classification, respectively. Full article

In an era dominated by Internet of Things (IoT) devices, software-as-a-service (SaaS) platforms, and rapid advances in cloud and edge computing, the demand for efficient and lightweight models suitable for resource-constrained devices such as data processing units (DPUs) has surged. Traditional deep learning models, such as convolutional neural networks (CNNs), pose significant computational and memory challenges, limiting their use in resource-constrained environments. Echo State Networks (ESNs), based on reservoir computing principles, offer a promising alternative with reduced computational complexity and shorter training times. This study explores the applicability of ESN-based architectures in image classification and weather forecasting tasks, using benchmarks such as the MNIST, FashionMnist, and CloudCast datasets. Through comprehensive evaluations, the Multi-Reservoir ESN (MRESN) architecture emerges as a standout performer, demonstrating its potential for deployment on DPUs or home stations. In exploiting the dynamic adaptability of MRESN to changing input signals, such as weather forecasts, continuous on-device training becomes feasible, eliminating the need for static pre-trained models. Our results highlight the importance of lightweight models such as MRESN in cloud and edge computing applications where efficiency and sustainability are paramount. This study contributes to the advancement of efficient computing practices by providing novel insights into the performance and versatility of MRESN architectures. By facilitating the adoption of lightweight models in resource-constrained environments, our research provides a viable alternative for improved efficiency and scalability in modern computing paradigms. Full article

In the context of TinyML, many research efforts have been devoted to designing forward topologies to support On-Device Learning. Reaching this target would bring numerous advantages, including reductions in latency and computational complexity, stronger privacy, data safety and robustness to adversarial attacks, higher resilience against concept drift, etc. However, On-Device Learning on resource constrained devices poses severe limitations to computational power and memory. Therefore, deploying Neural Networks on tiny devices appears to be prohibitive, since their backpropagation-based training is too memory demanding for their embedded assets. Using Extreme Learning Machines based on Convolutional Neural Networks might be feasible and very convenient, especially for Feature Extraction tasks. However, it requires searching for a randomly initialized topology that achieves results as good as those achieved by the backpropagated model. This work proposes a novel approach for automatically composing an Extreme Convolutional Feature Extractor, based on Neural Architecture Search and Bayesian Optimization. It was applied to the CIFAR-10 and MNIST datasets for evaluation. Two search spaces have been defined, as well as a search strategy that has been tested with two surrogate models, Gaussian Process and Random Forest. A performance estimation strategy was defined, keeping the feature set computed by the MLCommons-Tiny benchmark ResNet as a reference model. In as few as 1200 search iterations, the proposed strategy was able to achieve a topology whose extracted features scored a mean square error equal to 0.64 compared to the reference set. Further improvements are required, with a target of at least one order of magnitude decrease in mean square error for improved classification accuracy. The code is made available via GitHub to allow for the reproducibility of the results reported in this paper. Full article

Sensor-based human activity recognition is becoming ever more prevalent. The increasing importance of distinguishing human movements, particularly in healthcare, coincides with the advent of increasingly compact sensors. A complex sequence of individual steps currently characterizes the activity recognition pipeline. It involves separate data collection, preparation, and processing steps, resulting in a heterogeneous and fragmented process. To address these challenges, we present a comprehensive framework, HARE, which seamlessly integrates all necessary steps. HARE offers synchronized data collection and labeling, integrated pose estimation for data anonymization, a multimodal classification approach, and a novel method for determining optimal sensor placement to enhance classification results. Additionally, our framework incorporates real-time activity recognition with on-device model adaptation capabilities. To validate the effectiveness of our framework, we conducted extensive evaluations using diverse datasets, including our own collected dataset focusing on nursing activities. Our results show that HARE’s multimodal and on-device trained model outperforms conventional single-modal and offline variants. Furthermore, our vision-based approach for optimal sensor placement yields comparable results to the trained model. Our work advances the field of sensor-based human activity recognition by introducing a comprehensive framework that streamlines data collection and classification while offering a novel method for determining optimal sensor placement. Full article

Determining the cyclic-alternating-pattern (CAP) phases in sleep using electroencephalography (EEG) signals is crucial for assessing sleep quality. However, most current methods for CAP classification primarily rely on classical machine learning techniques, with limited implementation of deep-learning-based tools. Furthermore, these methods often require manual feature extraction. Herein, we propose a fully automatic deep-learning-based algorithm that leverages convolutional neural network architectures to classify the EEG signals via their time-frequency representations. Through our investigation, we explored using time-frequency analysis techniques and found that Wigner-based representations outperform the commonly used short-time Fourier transform for CAP classification. Additionally, our algorithm incorporates contextual information of the EEG signals and employs data augmentation techniques specifically designed to preserve the time-frequency structure. The model is developed using EEG signals of healthy subjects from the publicly available CAP sleep database (CAPSLPDB) on Physionet. An experimental study demonstrates that our algorithm surpasses existing machine-learning-based methods, achieving an accuracy of 77.5% on a balanced test set and 81.8% when evaluated on an unbalanced test set. Notably, the proposed algorithm exhibits efficiency and scalability, making it suitable for on-device implementation to enhance CAP identification procedures. Full article