Search Results (147)

Background: The autonomic nervous system (ANS) in individuals with obsessive–compulsive disorder (OCD) is often characterized by heightened sympathetic tone and reduced parasympathetic flexibility. This exploratory pilot study aimed to provide a detailed, dynamic portrait of autonomic dysregulation. Methods: Heart rate variability (HRV), galvanic skin response (GSR), and oculomotor measures (pupil size, fixations, saccade velocity) were recorded in 31 participants with OCD (obsessive–compulsive disorder) and 46 healthy controls while viewing 18 videos, half of them designed to trigger specific OCD dimensions and half neutral or positive. Results: The OCD group exhibited a state of sympathetic hyperactivation, evidenced by significantly lower mean NN intervals (average interval between normal heartbeats) and a distinct GSR (galvanic skin response) pattern marked by a rising tonic component across the experiment. Healthy controls exhibited increases in SDNN (standard deviation of NN intervals) and RMSSD (root mean square of successive differences) following pleasant videos, whereas this response was less pronounced and emerged later in the OCD group, suggesting reduced autonomic flexibility. Oculomotor data revealed faster saccade velocities during video viewing in the OCD group; however, unlike controls, their eye movements did not correlate with the pleasantness of the content. Conclusions: These results support the theory that individuals with OCD may exhibit greater sympathetic activation and delayed parasympathetic engagement. The findings suggest that autonomic inflexibility may be a core, measurable feature of OCD. Full article

Generating early and reliable fire/smoke alarms from real-world video surveillance remains challenging because visually ambiguous patterns such as sunlight, reflections, clouds, mist, steam, and illumination changes often trigger unstable frame-level false alarms. This paper presents the Spatio-Temporal Verification Network for Fire and Smoke Alarm Validation (STV-FSANet), a detect–track–verify framework that first localizes candidate fire/smoke regions, associates them into temporal tracks, and then verifies the observed sequence of each active track online as fire, smoke, or false fire/smoke. The verifier combines a primary appearance stream from cropped candidate regions with lightweight geometric cues derived from bounding-box position, scale, motion, and short-term fluctuation. A dual-branch GRU models long-term track history, while recent temporal pooling emphasizes newly observed evidence for streaming decisions. To support temporal learning, we construct the Fire–Smoke Alarm Verification (FSAV) Tracklet Dataset from 1347 source videos, yielding 58,733 parent tracks and 2.06 million annotated track frames. The best matched-context STV-FSANet achieves 97.09% test accuracy and 95.81% macro-F1, and both shorter-context models are within 0.5 percentage points of their final prefix metrics by 1.0 s. The results indicate that compact temporal evidence is more useful than simply accumulating longer 96-frame histories. The proposed model also rejects false fire/smoke tracklets with 90.2% recall, demonstrating the value of explicit hard-negative modeling, while the decoupled design allows the verifier to be reused with future detector backbones. TensorRT FP16 deployment reaches 109.55 frames/s on an RTX 4060 Laptop GPU and 39.51 frames/s on a Jetson AGX Orin DevKit. Full article

This paper addresses the urgent scientific and applied problem of automatic tracking of highly maneuverable Unmanned Aerial Vehicles (UAVs) using systems based on platforms with limited computing power (Edge Computing). The paper analyzes the shortcomings of classical correlation trackers and detectors based on deep neural networks when tracking targets with non-linear trajectories. A hybrid tracking method is proposed, combining the speed of a Kernelized Correlation Filter (KCF) and the accuracy of a neural network detector (YOLO11s). A key feature of the method is the developed algorithm for adaptive Kalman Filter correction, which utilizes a dynamic, scale-invariant Prediction Error metric as a trigger for motion anomaly detection. This allows the system to distinguish between measurement noise and sharp target maneuvers, executing an adaptive state reset using finite differences only at critical moments. Experimental validation on edge hardware (Raspberry Pi 5) using highly dynamic video sequences from the UAV123 and VisDrone datasets demonstrated that the proposed approach maintains an average processing speed of 18.89 FPS. By limiting deep neural network invocations to merely 2.71% of total frames, the algorithm successfully curtails thermal throttling while achieving a global Mean Root Square Error (RMSE) of 259.10 pixels across highly erratic trajectories. The method ensures high tracking reliability without a critical increase in computational load, making it highly suitable for use in autonomous embedded systems. Full article

Real-world home monitoring requires sensing systems that can capture daily behaviour without continuous raw-video retention or excessive user burden. However, domestic environments present irregular activity timing, fragmented human presence, asynchronous multimodal events, and privacy-sensitive data management. This study proposes an event-driven multimodal sensing and computing framework for context-aware home monitoring using stereo vision and dietary event anchoring. The framework integrates stereo RGB-based three-dimensional human motion sensing, dining-zone-triggered meal image acquisition, runtime event orchestration, timestamp-based cross-modal synchronization, privacy-aware local storage, and large-language-model-assisted dietary context interpretation. Instead of continuously recording all sensor streams, the system activates and organizes sensing through human presence detection, debounce logic, cooldown-based session control, and dining-zone occupancy events. Meal-related events are used as contextual anchors to associate motion sessions and dietary observations into synchronized behavioural episodes. The prototype was deployed for 11 consecutive days in a real kitchen–dining environment, with the stabilized real-time monitoring phase evaluated from 11 to 14 February 2026. During this phase, the system generated 26 event-driven motion sessions and 51,165 captured pose frames, of which 25,925 were valid. Sustained active sessions accounted for 30.8% of all sessions but contributed 81.5% of captured pose frames, indicating that event-driven orchestration concentrated motion data within behaviourally meaningful activity windows. Eight meal-related records were obtained, seven of which overlapped with motion sessions, resulting in 87.5% meal-event overlap coverage. Structured pose outputs required approximately 550 kB/min, corresponding to about 33 MB/h of recorded pose data. LLM-assisted meal-image interpretation achieved a mean absolute percentage error of 25.44%, supporting its use for coarse dietary-context description rather than precise nutritional quantification. However, this result is interpreted only as evidence for coarse dietary-context description and not as validation of a precise nutritional or clinical dietary assessment method. These results demonstrate the system-level feasibility of transforming irregular domestic observations into structured, temporally indexed, and privacy-aware multimodal behavioural records for future home monitoring applications. Full article

Due to enclosed space, long evacuation distances, and complex path structures, key nodes in deep underground spaces are prone to forming bottlenecks during fire evacuation. To collect evacuation behavior data at potential bottlenecks, an interactive video-based hypothetical choice (HC) experiment was conducted with 104 valid samples. Exit distance, sub-safe zone setting, congestion, pedestrian flow guidance, and smoke were systematically examined. The results showed that: (a) exit distance, sub-safe zone setting, congestion at the nearest exit, and smoke significantly affected evacuation decisions, with clear avoidance of near-exit congestion and smoke; (b) congestion on paths to non-nearest exits had a relatively weak effect, and pedestrian flow guidance did not produce significant herding; and (c) gender, age, professional background, and evacuation experience influenced path choice differences under certain conditions. Notably, evacuees prioritized smoke avoidance over all other cues, while congestion triggered non-compensatory route switching rather than herding behavior. These findings enrich the empirical database on pedestrian evacuation dynamics in deep underground spaces and provide a quantitative basis for evacuation simulation, spatial optimization, and safety management. Full article

Adaptive bitrate streaming over HTTP relies on carefully constructed bitrate ladders and ordered sets of bitrate–resolution pairs to deliver optimal perceptual quality under fluctuating network conditions. While content-aware methods based on convex hull optimisation have substantially improved ladder efficiency for Video-on-Demand, they require exhaustive multi-resolution pre-encoding that is computationally prohibitive under the real-time constraints of live streaming. This challenge is compounded by the H.266/Versatile Video Coding (VVC) standard, which offers approximately 50% compression gains over HEVC at 8–10× the encoding complexity. This paper presents LiveCH-VVC, a latency-aware dynamic bitrate ladder prediction framework for VVC-encoded live streaming over Low-Latency DASH (LL-DASH) with CMAF packaging. The framework introduces four integrated modules: (i) a Lightweight Dual-Path CNN (LDP-CNN), obtained via teacher–student knowledge distillation (∼5 M parameters, 148 ms GPU inference), that jointly extracts spatial–temporal features from raw frames and compression-domain statistics from a fast VVC probe encode; (ii) an adaptive scene change detector with exponential moving average thresholding (F1 = 0.925) that triggers ladder updates only upon significant complexity shifts; (iii) a temporally augmented XGBoost multi-label classifier that predicts latency-constrained Pareto-optimal bitrate–resolution pairs; and (iv) an online adaptation engine that integrates Common Media Client Data (CMCD) feedback from CDN edge servers for continuous closed-loop refinement. Comprehensive evaluation on 81 UHD sequences (∼4050 CMAF segments) from three benchmark datasets demonstrates an average BD-Rate of $+0.68\%$ relative to the per-segment oracle convex hull $5.4\times$ better than the state-of-the-art ARTEMIS framework ($+3.67\%$) while achieving 73.3% encoding time savings, 2.37 s end-to-end latency, and a QoE score of 81.6 in live simulation with 100 concurrent clients. Ablation analysis confirms that the dual-path compression-domain branch ($+0.44$ pp) and temporal context augmentation ($+0.35$ pp) are the primary performance drivers, while the online adaptation mechanism provides $42\%$ relative improvement over extended streaming sessions. Full article

Drone-based shark surveillance has been implemented as a non-lethal mitigation method to minimise the risk of human–shark interactions along beaches of New South Wales (NSW), Australia. However, real-time misidentification remains problematic, often triggering unnecessary countermeasures due to marine animals that pose little to no risk to humans. We investigated shark misidentification in drone surveys by comparing real-time identification with post-flight verification across 900 flights. Post-flight analyses revealed false-positive detection rates of 53%, 79%, and 100% for bull (Carcharhinus leucas), white (Carcharodon carcharias), and tiger (Galeocerdo cuvier) sharks, respectively, which collectively are the ‘target’ sharks of mitigation measures in NSW. Of the 269 flights in which sharks were identified in real time as target sharks, 62% were confirmed post-flight as other sharks (i.e., whaler species, grey nurse, leopard, or wobbegong), sharks that could not be identified (unknown sharks), or non-shark species (i.e., guitarfish). Conversely, 25% of flights with target sharks identified post-flight were recorded in real time as ‘other’ or ‘unknown’ sharks. Overall, real-time classification overestimated the presence of target sharks, with an apparent prevalence approximately twice the true prevalence. Countermeasure activations based on real-time classification of target sharks were accurate in only 36% of instances. Non-shark species (i.e., guitarfish or gamefish) also triggered 39 countermeasures, including 28 water evacuations. Integrating artificial intelligence or other advances (e.g., higher-resolution video on larger screens) may enhance the effectiveness of drone-based surveillance by assisting pilots with real-time shark detection and identification. Full article

Countermovement jump (CMJ) analysis is widely used to assess lower limb neuromuscular function, but commercial force plates often suffer from high cost, closed algorithms, and lack of bilateral independent measurement. This study developed and evaluated a dual channel strain gauge force plate system featuring open architecture and hardware-triggered video synchronization. The system consists of two physically isolated plates, each with four full bridge strain beams, a precision analog front end, and a 2000 Hz acquisition unit. A microcontroller-based hardware trigger synchronizes force data with video capture. Custom host software implements adaptive jump phase recognition and calculates peak force (PF), concentric impulse, jump height, rate of force development (RFD), and asymmetry index (ASI). Validation included static mass measurements in 14 participants, low-load static calibration (5.0–30.0 kg), free-fall impulse validation (7.00 to 31.32 N·s), 240 fps high-speed video cross validation of flight time, ecological-validity comparison with published AMTI-based force-plate data, and 48 h test–retest reliability assessment. Static mass measurement showed a mean absolute percentage error (MAPE) of 1.01% and a coefficient of determination (${\mathrm{R}}^2$) of 0.9992, while low-load testing confirmed excellent linearity (${\mathrm{R}}^2>0.996$) and minimal absolute error (mean absolute error = 0.34 kg) at lighter weights. Dynamic impulse validation yielded ${\mathrm{R}}^2>0.997$ and MAPE < 3%. Flight time agreement with high-speed video was within ±10 ms. Test–retest reliability was excellent for concentric impulse (intraclass correlation coefficient (ICC) = 0.997) and jump height (ICC = 0.987), and good for PF (ICC = 0.962) and rate of force development at 100 ms (RFD_100ms) (ICC = 0.883). The physically isolated dual-plate architecture effectively captured bilateral force differences, although the ASI demonstrated moderate reliability (ICC = 0.748), likely reflecting the inherent biological variability in bilateral coordination. The ecological-validity comparison further indicated that the macroscopic kinetic outputs of the proposed system fell within the expected physiological and biomechanical ranges reported for adult CMJ testing. Overall, these findings support the study hypothesis that the proposed dual-channel force plate system provides a valid, reliable, and cost-effective solution for synchronized bilateral CMJ kinetic assessment in sports performance monitoring and biomechanical research, while offering improved accessibility through an open-source and transparent analysis framework with a hardware cost below 500 USD. Full article

Wave overtopping is one of the most complex coastal hazards to characterize in field conditions due to its high non-linearity and the interaction between unsteady hydrodynamics and wave–structure processes. To get insights into the underlying occurrence and persistence of overtopping, this study proposes an integration of numerical and data-driven models. Multi-month field observations made at a breakwater are used to investigate the hydro-meteorological parameters causing overtopping initiation and persistence. High-frequency video-derived overtopping detections are combined with coupled ADCIRC–UnSWAN (ADvanced CIRCulation–Unstructured Simulating WAves Nearshore) hindcasts to construct near-structure hydro-meteorological conditions. The results reveal a clear dynamical asymmetry showing that overtopping initiation corresponds to exceedance of crest elevation at individual wave-scale associated with elevated wave height, water level, wave steepness, and wind characteristics, whereas overtopping persistence depends on short-term temporal effects associated with wave energy, direction, and sustained water levels. Gradient-boosted decision trees, temporal convolutional networks, and Transformer models are employed, demonstrating that persistence cannot be inferred from instantaneous sea-states alone, indicating a separation of timescales between triggering and sustained overtopping dynamics. These findings provide field-scale evidence of distinct hydrodynamic regimes governing overtopping processes, highlighting the importance of temporal characteristics for understanding overtopping dynamics and developing predictive coastal hazard frameworks. Full article

Student mental health and academic stress are increasingly addressed through digital monitoring, yet evidence for personalized physiological thresholds based on exhaled VOCs, their in-the-wild feasibility, and their trigger–experience correspondence in everyday student life remains limited. This study examines whether exhaled breath signals can support personalized, real-world stress monitoring in university students using a two-stage design that moves from laboratory calibration to daily life validation. A total of 24 university students took part in the laboratory phase (Study 1; N = 24). Under two stress tasks, a social-conflict video task and a Stroop task, we derived an individualized breath-trigger threshold (θ_i) for each participant. We then invited 21 of them to join a three-day field deployment (Study 2; N = 21). Each participant’s θ_i from Study 1 was used directly as the trigger threshold for daily monitoring in order to test the association between trigger events and subjectively noticeable emotional deviations and to assess preliminary trigger–experience correspondence in daily life. The results show that 78.6% of paired trigger–EMA records were rated as subjectively salient, with 93.9% of these rated at medium-to-high intensity. These events occurred most frequently during study/work activities (60.6%), in dorm/home settings (57.6%), and when participants were alone (63.6%), suggesting that the triggers captured personally meaningful emotional episodes embedded in routine academic life rather than random physiological fluctuations. Overall, this study presents a portable breath-based emotion sampling device for student academic contexts and a reproducible protocol that combines laboratory thresholding with daily life validation. The findings provide preliminary and exploratory indications of the feasibility and within-person transferability of VOC-based emotion detection in students, and offer methodological support for future digital emotion monitoring and intervention design based on breath signals. Full article

In the era of 5G and emerging 6G service-based architectures, research infrastructures require versatile, scalable tools for performance validation. This article presents an original distributed application-layer traffic generation system developed within the PL-5G National Laboratory for Advanced 5G Research. While dedicated hardware generators offer high precision, their prohibitive costs and rigid architectures often limit the scope of distributed experimental research. Shifting the testing paradigm to application-layer microservice interactions, our solution leverages general-purpose computing resources and a containerized microservice architecture to enable realistic, low-cost performance assessment. The primary objective of this study was to analyze the complex relationship between computing resource consumption and traffic generation efficiency. We conducted scalability experiments simulating diverse 5G/6G use cases, such as high-frequency Internet of Things (IoT) sensor data and real-time video streaming. Experimental results demonstrate a near-perfect linear relationship between CPU utilization and throughput for heavy workloads. In contrast, high-frequency packets trigger a critical exception, shifting the bottleneck to severe throughput saturation under intense request rates. The study concludes that the proposed architectural approach provides a flexible, cost-effective alternative to hardware-centric solutions. By identifying these hardware–software dependencies, the system enables efficient, scalable testing without specialized, expensive infrastructure. Full article

Fertilizer reduction and improved fertilization management are important approaches to achieving green and organic grape production and vineyard management. This study aims to examine whether digital agricultural extension tools can facilitate fertilizer reduction among grape growers, or whether their actual effects may be weakened by limited attention and information overload. Using data from a randomized controlled trial (RCT) involving 475 grape growers in Liaoning Province, this study employs ordinary least squares (OLS) models and propensity score matching (PSM) to test the effect of digital information loading on growers’ fertilizer-reduction behavior. Specifically, this study constructs three dimensions, including information loading breadth (comprehensiveness), loading depth (accuracy), and loading length (completeness), and employs ordered logit models and threshold regression models to explore their mechanism effects and threshold effects on fertilizer reduction. The results indicate that (1) digital information loadingsignificantly promotes fertilizer reduction behaviors among grape growers. (2) The breadth and depth of information loading facilitate fertilizer reduction by enhancing growers’ perceived usefulness, whereas this mechanism is not significant for the length dimension. (3) Information loading length has a threshold effect on fertilizer reduction behavior, which means there is an optimal time interval for single video viewing; beyond that, the limited attention triggered information overload that causes the fertilizer reduction behavior to disappear. (4) There are also long-term effects of digital information loading. These findings provide micro-level empirical evidence for promoting green and organic-oriented grape production systems, and offer policy implications for countries pursuing sustainable agricultural transition and fertilizer reduction. Full article

Shared immersive environment sports venues, virtual classrooms, and collaborative workspaces require multiple users to stream 360° videos simultaneously over the same edge network, yet every existing adaptive bitrate system optimises each viewer in isolation. This self-interested behaviour triggers a bandwidth auction that chronically starves the most uncertain viewers: Jain’s Fairness Index for ten independently optimised agents routinely falls below 0.85. We present FairEdge360, a hierarchical multi-agent reinforcement learning framework that reformulates multi-user 360° streaming as a Decentralised Partially Observable Markov Decision Process (Dec-POMDP) and proves, formally, that fairness and quality are complementary rather than competing objectives. Three tightly coupled innovations make this possible. First, a Lightweight Uncertainty Estimator (LUE) a compact 8385-parameter four-layer MLP evaluates per-device viewport prediction confidence $c_t^i=\sigma \left(w_4^{\top }{h}_3\right)$ in under approximately 2.1 ms on commodity smartphones (95th percentile, iPhone 12 A14 Bionic), enabling selective edge offloading that reduces device energy consumption by 38.9%. Second, a variational Graph Neural Network compresses each agent’s 256-dimensional GRU state into a 32-byte INT8 latent, transmitted over a dynamic RTT-gated neighbourhood graph at 96 bytes per agent per 500 ms 75% less overhead than competing approaches. Third, the edge coordinator maximises the Nash social welfare objective $\mathrm{NSW}={\left({\prod }_{i=1}^N{Q}_i\right)}^{1/N}$, whose gradient $\partial \mathrm{NSW}/\partial Q_i\propto 1/Q_i$ automatically prioritises the most disadvantaged viewer; a formal proof guarantees that every Pareto-optimal policy satisfies $Q_i/{\sum }_j{Q}_j\ge 1/N$. Counterfactual advantage estimation correctly attributes each agent’s marginal contribution to the global reward, eliminating the credit-assignment ambiguity inherent in standard multi-agent baselines. Evaluated on 284 users, 52 omnidirectional videos, and 10,000 real network traces spanning 4G LTE, 5G mmWave, HSDPA, and campus WiFi, FairEdge360 raises Jain’s Fairness Index from 0.934 to 0.976 (+4.5%), improves worst-case user quality-of-experience from MOS 2.54 to MOS 3.21 (+26.4%), and halves rebuffering rate from 2.1% to 1.1%, all within a 20 ms motion-to-photon budget on a commodity smartphone. Full article

High-density urban plazas hosting multi-session public events often experience pulsed inflows, prolonged crowd retention, and localized bottleneck congestion, creating crowd-safety risks that cannot be fully captured by static capacity or total evacuation time alone. This study develops an agent-based simulation framework to evaluate ultra-high-density exposure and evacuation dynamics in Guangzhou Huacheng Plaza during the International Light Festival. The model was constructed in AnyLogic using site-layout data, event organization records, official attendance information, historical event timelines, and publicly available video observations. Two scenarios were examined: normal dynamic entry–exit operation under different inter-performance intervals, and overload-triggered evacuation under alternative spatial management strategies. Model calibration and event-process validation were conducted by comparing simulated congestion hotspots, key event timing, delayed dispersal patterns, and evacuation-duration ranges with historical observations and documented event records. The results show that extending the inter-performance interval from 90 min to 120 min reduced the overload duration from 75 min to 5 min and decreased cumulative ultra-high-density exposure from 25.62 to 13.93. Under overload evacuation, zonal guidance mainly improved early-stage crowd redistribution, whereas increased exit capacity produced a stronger reduction in total evacuation time and sustained congestion. Total evacuation time decreased from 185 min in the baseline condition to 160 min under the combined strategy, while effective discharge capacity increased from 231.12 to 338.28 pedestrians/min. These findings indicate that crowd safety in open urban plazas depends not only on total attendance, but also on event pacing, bottleneck recovery time, and effective discharge capacity. The proposed exposure-oriented framework provides a quantitative basis for evaluating crowd accumulation and evacuation strategies in high-density open public spaces. Full article

In the digital economy era, Urban vitality has transitioned into an intertwined Virtual-Physical system. This study examines Changsha’s five urban districts through a dual-dimensional framework bridging physical (social, economic, cultural, and ecological) and virtual (video, social, and digital life) dimensions. Integrating Coupling Coordination Degree (CCD) and XGBoost-SHAP models, we elucidate the spatial patterns and nonlinear drivers of Virtual-Physical synergy. The results indicate that: (1) Urban Vitality exhibits a significant center-periphery gradient. Although the Coupling Degree between the two dimensions is high, the overall CCD remains relatively low, reflecting pervasive spatial mismatches. Notably, 55 units display a reverse pattern where Virtual Vitality surpasses Physical Vitality, suggesting that digital flows can reconfigure urban space by transcending traditional locational constraints. (2) Interactions within the built environment exert pronounced threshold effects. Structural elements require specific critical masses to activate synergy, beyond which marginal returns diminish, as exemplified by the U-shaped effect of the Green View Index and the inverted U-shaped effect of Spatial Enclosure on CCD. (3) Interaction analysis identifies building density as a multiplier, unlocking the synergistic potential of land-use mix and transport networks once critical thresholds are surpassed. Furthermore, the efficacy of population and transit relies on dense road networks and intersection, while functional diversity buffers against negative micro-environmental impacts. This study advocates for a shift from facility-increment to threshold-triggered precision strategies in urban regeneration, providing empirical support for human-centric planning in the digital twin era. Full article