Real-time sensor-based applications in pervasive computing require edge-deployable models to ensure low latency, privacy, and efficient interaction. A prime example is sensor-based human activity recognition (HAR), where models must balance accuracy with stringent resource constraints. Yet many deep learning approaches treat temporal sensor signals as black-box sequences, overlooking spectral–temporal structure while demanding excessive computation. We present SPECTRA, a deployment-first, co-designed spectral–temporal architecture that integrates short-time Fourier transform (STFT) feature extraction, depthwise separable convolutions, and channel-wise self-attention to capture spectral–temporal dependencies under real edge runtime and memory constraints. A compact bidirectional GRU with attention pooling summarizes within-window dynamics at low cost, reducing downstream model burden while preserving accuracy. Across five public HAR datasets, SPECTRA matches or approaches larger CNN/LSTM/Transformer baselines while substantially reducing parameters, latency, and energy. Deployments on a Google Pixel 9 smartphone and an STM32L4 microcontroller further demonstrate end-to-end deployable real-time, private, and efficient HAR.

E-textiles are rapidly emerging as an important area of electronic circuit applications. It also facilitates many socially important applications such as personalized health, elderly care, and smart agriculture. However, the environmental impact and sustainability of e-textiles remain very problematic. STELEC, short for Sustainable Textile ELECtronics, is an interdisciplinary research project funded by the European Innovation Council (EIC) under the Pathfinder programme on the responsible elec-tronics topic seeking cutting-edge innovation. STELEC started in September 2024 and is in its initial stage. The project is a multinational collaboration of research institutes, universities and companies across Europe. It aims at developing next-generation textile-based electronics in applications from sensing, processing to AI, with a commitment to full lifecycle sustainability.

Understanding human-to-human interactions, especially in contexts like public security surveillance, is critical for monitoring and maintaining safety. Traditional activity recognition systems are limited by fixed vocabularies, predefined labels, and rigid interaction categories that often rely on choreographed videos and overlook concurrent interactive groups. These limitations make such systems less adaptable to real-world scenarios, where interactions are diverse and unpredictable. In this paper, we propose an open vocabulary human-to-human interaction recognition (OV-HHIR) framework that leverages large language models to generate open-ended textual descriptions of both seen and unseen human interactions in open-world settings without being confined to a fixed vocabulary. Additionally, we create a comprehensive, large-scale human-to-human interaction dataset by standardizing and combining existing public human interaction datasets into a unified benchmark. Extensive experiments demonstrate that our method outperforms traditional fixed-vocabulary classification systems and existing cross-modal language models for video understanding, setting the stage for more intelligent and adaptable visual understanding systems in surveillance and beyond.

Breathing is a spontaneous but controllable body function that can be used for hands-free interaction. Our work introduces “iBreath”, a novel system to detect breathing gestures similar to clicks using bio-impedance. We evaluated iBreath’s accuracy and user experience using two lab studies (n=34). Our results show high detection accuracy (F1-scores > 95.2%). Furthermore, the users found the gestures easy to use and comfortable. Thus, we developed eight practical guidelines for the future development of breathing gestures. For example, designers can train users on new gestures within just 50 seconds (five trials), and achieve robust performance with both user-dependent and user-independent models trained on data from 21 participants, each yielding accuracies above 90%. Users preferred single clicks and disliked triple clicks. The median gesture duration is 3.5-5.3 seconds. Our work provides solid ground for researchers to experiment with creating breathing gestures and interactions.

We present a novel local-global feature fusion framework for body-weight exercise recognition with floor-based dynamic pressure maps. One step further from the existing studies using deep neural networks mainly focusing on global feature extraction, the proposed framework aims to combine local and global features using image processing techniques and the YOLO object detection to localize pressure profiles from different body parts and consider physical constraints. The proposed local feature extraction method generates two sets of high-level local features consisting of cropped pressure mapping and numerical features such as angular orientation, location on the mat, and pressure area. In addition, we adopt a knowledge distillation for regularization to preserve the knowledge of the global feature extraction and improve the performance of the exercise recognition. Our experimental results demonstrate a notable 11 percent improvement in F1 score compared to the baseline 3DCNN for exercise recognition while preserving label-specific features.

Automatic and precise fitness activity recognition can be beneficial in aspects from promoting a healthy lifestyle to personalized preventative healthcare. While IMUs are currently the prominent fitness tracking modality, through iMove, we show bio-impedence can help improve IMU-based fitness tracking through sensor fusion and contrastive learning. To evaluate our methods, we conducted an experiment including six upper body fitness activities performed by ten subjects over five days to collect synchronized data from bio-impedance across two wrists and IMU on the left wrist. The contrastive learning framework uses the two modalities to train a better IMU-only classification model, where bio-impedance is only required at the training phase, by which the average Macro F1 score with the input of a single IMU was improved by 3.22 % reaching 84.71 % compared to the 81.49 % of the IMU baseline model. We have also shown how bio-impedance can improve human activity recognition (HAR) directly through sensor fusion, reaching an average Macro F1 score of 89.57 % (two modalities required for both training and inference) even if Bio-impedance alone has an average macro F1 score of 75.36 %, which is outperformed by IMU alone. In addition, similar results were obtained in an extended study on lower body fitness activity classification, demonstrating the generalisability of our approach.Our findings underscore the potential of sensor fusion and contrastive learning as valuable tools for advancing fitness activity recognition, with bio-impedance playing a pivotal role in augmenting the capabilities of IMU-based systems.

Manufacturing industries strive to improve production efficiency and product quality by deploying advanced sensing and control systems. Wearable sensors are emerging as a promising solution for achieving this goal, as they can provide continuous and unobtrusive monitoring of workers’ activities in the manufacturing line. This article presents a novel wearable sensing prototype that combines IMU and body capacitance sensing modules to recognize worker activities in the manufacturing line. To handle these multimodal sensor data, we propose and compare early, and late sensor data fusion approaches for multichannel time-series convolutional neural networks and deep convolutional LSTM. We evaluate the proposed hardware and neural network model by collecting and annotating sensor data using the proposed sensing prototype and Apple Watches in the testbed of the manufacturing line. Experimental results demonstrate that our proposed methods achieve superior performance compared to the baseline methods, indicating the potential of the proposed approach for real-world applications in manufacturing industries. Furthermore, the proposed sensing prototype with a body capacitive sensor (BCS) and feature fusion method improves by 6.35%, yielding a 9.38% higher macro F1 score than the proposed sensing prototype without a BCS and Apple Watch data, respectively.

Distractions caused by digital devices are increasingly causing dangerous situations on the road, particularly for more vulnerable road users like cyclists. While researchers have been exploring ways to enable richer interaction scenarios on the bike, safety concerns are frequently neglected and compromised. In this work, we propose Head 'n Shoulder, a gesture-driven approach to bike interaction without affecting bike control, based on a wearable garment that allows hands- and eyes-free interaction with digital devices through integrated capacitive sensors. It achieves an average accuracy of 97% in the final iteration, evaluated on 14 participants. Head 'n Shoulder does not rely on direct pressure sensing, allowing users to wear their everyday garments on top or underneath, not affecting recognition accuracy. Our work introduces a promising research direction: easily deployable smart garments with a minimal set of gestures suited for most bike interaction scenarios, sustaining the rider's comfort and safety.

In this work, we propose an open-source scalable end-to-end RTL framework FieldHAR, for complex human activ-ity recognition (HAR) from heterogeneous sensors using artificial neural networks (ANN) optimized for FPGA or ASIC integration. FieldHAR aims to address the lack of apparatus to transform complex HAR methodologies often limited to offline evaluation to efficient runtime edge applications. The framework uses parallel sensor interfaces and integer-based multi-branch convolutional neural networks (CNNs) to support flexible modality extensions with synchronous sampling at the maximum rate of each sensor. To validate the framework, we used a sensor-rich kitchen scenario HAR application which was demonstrated in a previous offline study. Through resource-aware optimizations, with FieldHAR the entire RTL solution was created from data acquisition to ANN inference taking as low as 25% logic elements and 2% memory bits of a low-end Cyclone IV FPGA and less than 1% accuracy loss from the original FP32 precision offline study. The RTL implementation also shows advantages over MCU-based solutions, including superior data acquisition performance and virtually eliminating ANN inference bottleneck.

Early prediction of remaining useful life (RUL) is crucial for effective battery management across various industries, ranging from household appliances to large-scale applications. Accurate RUL prediction improves the reliability and maintainability of battery technology. However, existing methods have limitations, including assumptions of data from the same sensors or distribution, foreknowledge of the end of life (EOL), and neglect to determine the first prediction cycle (FPC) to identify the start of the unhealthy stage. This paper proposes a novel method for RUL prediction of Lithium-ion batteries. The proposed framework comprises two stages: determining the FPC using a neural network-based model to divide the degradation data into distinct health states and predicting the degradation pattern after the FPC to estimate the remaining useful life as a percentage. Experimental results demonstrate that the proposed method outperforms conventional approaches in terms of RUL prediction. Furthermore, the proposed method shows promise for real-world scenarios, providing improved accuracy and applicability for battery management.

This work presents an active software instrument allowing deep learning architects to interactively inspect neural network models' output behavior from user-manipulated values in any latent layer. Latent Inspector offers multiple dimension reduction techniques to visualize the model's high dimensional latent layer output in human-perceptible, two-dimensional plots. The system is implemented with Node.js front end for interactive user input and Python back end for interacting with the model. By utilizing a general and modular architecture, our proposed solution dynamically adapts to a versatile range of models and data structures. Compared to already existing tools, our asynchronous approach of separating the training process from the inspection offers additional possibilities, such as interactive data generation, by actively working with the model instead of visualizing training logs. Overall, Latent Inspector demonstrates the possibilities as well as the appearing limits for providing a generalized, tool-based concept for enhancing model insight in terms of explainable and transparent AI.

The estimation of 3D human body shape and clothing measurements is crucial for virtual try-on and size recommendation problems in the fashion industry but has always been a challenging problem due to several conditions, such as lack of publicly available realistic datasets, ambiguity in multiple camera resolutions, and the undefinable human shape space. Existing works proposed various solutions to these problems but could not succeed in the industry adaptation because of complexity and restrictions. To solve the complexity and challenges, in this paper, we propose a simple yet effective architecture to estimate both shape and measures from frontal- and side-view images. We utilize silhouette segmentation from the two multi-view images and implement an auto-encoder network to learn low-dimensional features from segmented silhouettes. Then, we adopt a kernel-based regularized regression module to estimate the body shape and measurements. The experimental results show that the proposed method provides competitive results on the synthetic dataset, NOMO-3d-400-scans Dataset, and RGB Images of humans captured in different cameras.

Current research in eSports lacks the tools for proper game practising and performance analytics. The majority of prior work relied only on in-game data for advising the players on how to perform better. However, in-game mechanics and trends are frequently changed by new patches limiting the lifespan of the models trained exclusively on the in-game logs. In this article, we propose the methods based on the sensor data analysis for predicting whether a player will win the future encounter. The sensor data were collected from ten participants in 22 matches in the League of Legends video game. We have trained machine learning models, including the transformer and gated recurrent unit, to predict whether the player wins the encounter taking place after some fixed time in the future. For 10-s forecasting horizon, the transformer neural network architecture achieves the ROC AUC score of 0.706. This model is further developed into the detector capable of predicting that a player will lose the encounter occurring in 10 s in 88.3% of cases with 73.5% accuracy. This might be used as a players’ burnout or fatigue detector, advising players to retreat. We have also investigated which physiological features affect the chance to win or lose the next in-game encounter.

We present the system CoRSA to incorporate integrated sensors in millimeter-scale packages for continuous cardiorespiratory (CR) evaluation in sports activities. CoRSA retrofits trending sports apparel to add on CR sensing capability. The system uses an air pressure sensor inside a vented mask to approximate a spirometer, and an earlobe pulse-oximeter (PO) to monitor heart rate (HR) and oxygen saturation (SpO2). CoRSA also includes an inertial measurement unit for tracking activity and future study on motion artifact correction on the CR signals in active sports. An aerobic exercise evaluation is also performed which shows results similar to sports studies using bulkier conventional medical equipment in the CR signals' characteristics.

We present LYRA, a modular in-flight system that enhances service and assists flight attendants during their work. LYRA enables passengers to browse and order services from their smartphones. Smart glasses and a smart shoe-clip with RFID reader module provides flight attendants with situated information. We gained first insights into how flight attendants and passengers use of the system during a long distance flight from Frankfurt to Houston.

Convolutional Neural Networks (CNNs) have become the state-of-the-art in various computer vision tasks, but they are still premature for most sensor data, especially in pervasive and wearable computing. A major reason for this is the limited amount of annotated training data. In this paper, we propose the idea of leveraging the discriminative power of pre-trained deep CNNs on 2-dimensional sensor data by transforming the sensor modality to the visual domain. By three proposed strategies, 2D sensor output is converted into pressure distribution imageries. Then we utilize a pre-trained CNN for transfer learning on the converted imagery data. We evaluate our method on a gait dataset of floor surface pressure mapping. We obtain a classification accuracy of 87.66%, which outperforms the conventional machine learning methods by over 10%.

We present a wearable textile sensor system for monitoring muscle activity, leveraging surface pressure changes between the skin and an elastic sport support band. The sensor is based on an 8×16 element fabric resistive pressure sensing matrix of 1cm spatial resolution, which can be read out with 50fps refresh rate. We evaluate the system by monitoring leg muscles during leg workouts in a gym out of the lab. The sensor covers the lower part of quadriceps of the user. The shape and movement of the two major muscles (vastus lateralis and medialis) are visible from the data during various exercises. The system registers the activity of the user for every second, including which machine he/she is using, walking, relaxing and adjusting the machines; it also counts the repetitions from each set and evaluate the force consistency which is related to the workout quality. 6 people participated in the experiment of overall 24 leg workout sessions. Each session includes cross-trainer warm-up and cool-down, 3 different leg machines, 4 sets on each machine. Plus relaxing, adjusting machines, and walking, we perform activity recognition and quality evaluation through 2-dimensional mapping and the time sequence of the average force. We have reached 81.7% average recognition accuracy on a 2s sliding window basis, 93.3% on an event basis, and 85.6% spotting F1-score. We further demonstrate how to evaluate the workout quality through counting, force pattern variation and consistency.

We present a novel sensor system for the support of nutrition monitoring. The system is based on smart table cloth equipped with a fine grained pressure textile matrix and a weight sensitive tablet. Unlike many other nutrition monitoring approaches, our system is unobtrusive, non privacy invasive and easily deployable in every day life. It allows the spotting and recognition of food intake related actions, such as cutting, scooping, stirring, etc., the identification of the plate/container on which the action is executed, and the tracking of the weight change in the containers. In other words, we can determine how many pieces are cut on the main dish plate, how many are taken from the side dish, how many sips are taken from the drink, how fast the food is being consumed and how much weight is taken overall. In addition, the distinction between different eating actions, such as cutting, scooping, poking, provides clues to the type of food taken and the way the meal is consumed. We have evaluated our system on 40 meals (5 subjects) in a real life living environment: for seven eating related actions (cutting, scooping, stirring, etc.), resulting in above 90% average recognition rate for person dependent cases, and spotting each action out of continuous data streams (average F1 score 87%).

There is a large class of routine physical exercises that are performed on the ground, often on dedicated "mats" (e.g. push-ups, crunches, bridge). Such exercises involve coordinated motions of different body parts and are difficult to recognize with a single body worn motion sensors (like a step counter). Instead a network of sensors on different body parts would be needed, which is not always practicable. As an alternative we describe a cheap, simple textile pressure sensor matrix, that can be unobtrusively integrated into exercise mats to recognize and count such exercises. We evaluate the system on a set of 10 standard exercises. In an experiment with 7 subjects, each repeating each exercise 20 times, we achieve a user independent recognition rate of 82.5% and a user independent counting accuracy of 89.9%. The paper describes the sensor system, the recognition methods and the experimental results.

We demonstrate through a pressure sensor matrix, that weight distribution on feet is influenced by body posture. A small cheap carpet equipped with low precision pressure sensor matrix is already sufficient to detect subtle activities and identity of the person on the carpet. By a 0.4 m2 matrix of 32 × 32, 12 bit pressure sensors, we achieve 78.7% accuracy for 11 test subjects performing 7 subtle activities (open 7 different drawers or cabinet doors) and 88.6% accuracy in recognizing who has performed the activities. We thus see the potential of using a single carpet as a unified approach in houses to detect how inhabitants interact with the furniture without attaching different sensors onto each single furniture.

In this paper we use ubiquitous smart-chairs to evaluate live presentations. We validate the hypothesis that the audiences' activities can be recognized with pressure sensors under chairs' legs (74.6% accuracy rate from 8 typical activities in 8 live presentations, each with 6 chairs seated), and ce