Mobile augmented reality (AR) enables visitors to access contextual digital information associated with physical locations and natural heritage resources. In protected natural areas, applications that can operate without continuous Internet access offer an alternative for delivering digital content independently of network availability. This study describes the development and evaluation of an offline mobile AR application for the Museum and Limpiopungo Lagoon circuits of Cotopaxi National Park, Ecuador. The application was built with Unity and Vuforia, stores its three-dimensional models on the device, and links predefined visual references to information about the surrounding environment. It was assessed through nine functional verification cases and a single-group pre–post study with 50 adult participants. All nine cases met their acceptance criteria. Scores on the AR and digital interpretation knowledge questionnaire rose from 53.3% to 88.5%, a mean paired difference of 35.2 percentage points (Wilcoxon signed-rank test, p<0.001). The mean System Usability Scale (SUS) score was 73.35. Positive responses to the nine visitor-experience items ranged from 64% to 82%, with the items analyzed individually because their thematic groups showed low internal consistency. The exploratory association between knowledge change and SUS was weak and not statistically significant (Spearman’s ρ=0.106, p=0.466). The findings document the implementation and evaluation of contextual mobile AR resources that can be accessed without continuous Internet connectivity in the selected visitor circuits. The increase in questionnaire scores should be interpreted cautiously because the study lacked a control group and the instrument did not directly assess biodiversity learning.
Automated interpretation of magnetic flux leakage (MFL) data is central to condition assessment of long-distance oil and gas pipelines. Yet, full-resolution panoramas are difficult to process without either suppressing small anomalies through resizing or fragmenting elongated objects at sliding-window boundaries. We propose MFL-ScrollNet, a recurrent sliding-window detector that formulates panoramic analysis as an ordered sequence problem. A gated recurrent unit propagates a compact hidden state across narrow axial windows, while geometric cross-window association in global panorama coordinates, complemented by learned novelty prediction, consolidates detections across adjacent windows. Experiments on the public PipeMFL-240K benchmark show that the full model attains an mAP50 of 0.535, an mAP50:95 of 0.332, a macro-averaged F1 score of 0.662, and a micro-averaged F1 score of 0.547. MFL-ScrollNet achieves competitive average-precision performance, surpassing the strongest reproduced static baseline, DINO-R50, while approaching the performance of the domain-specific STPDNet. At the selected operating point, MFL-ScrollNet achieves the highest macro- and micro-averaged F1 scores among the evaluated methods, demonstrating improved detection performance across both individual classes and the overall defect population. Controlled ablations show consistent gains from recurrent context, global-coordinate association, and learned novelty prediction. These results demonstrate that sequential cross-window reasoning improves both class-balanced and instance-level detection in panoramic MFL data.
Urban mobility systems are under increasing pressure from rapid urbanisation, population growth, and rising vehicle ownership, which contribute to recurrent congestion, unreliable travel times, environmental burdens, and declining service quality. Intelligent transportation systems (ITSs) support real-time sensing, communication, and traffic management, while digital twin (DT) technology provides continuously updated virtual representations that can support simulation, prediction, and decision support. In this study, a five-layer ITS–DT framework that links sensing and acquisition, communication and connectivity, data integration and digital twin synchronisation, analytics and intelligence, and operations and governance is developed. A structured review of literature published mainly between 2020 and 2025, a conceptual synthesis, and a prototype numerical experiment are undertaken. The proposed contribution is not the invention of the individual technologies, but their explicit cross-layer integration through a bidirectional data-and-control pipeline, interoperable interfaces, edge-to-cloud computing, and embedded governance. In a synthetic four-intersection corridor experiment with 50 paired stochastic replications, adaptive signal control reduced mean accrued queueing delay from 61.87 to 52.03 s per generated vehicle (15.9%) and the mean maximum approach queue from 84.28 to 29.52 vehicles (65.0%), while completed network exits increased by 0.7%. These prototype results demonstrate internal operational feasibility under stated assumptions rather than real-world city-scale validity. The framework therefore provides a transparent basis for the subsequent calibration, field validation, and context-specific implementation of integrated smart mobility systems.
Dynamic facial expression recognition (DFER) has advanced rapidly, but leading systems use transformer backbones with more than 80 million parameters and assume fully decoded, high-quality video. Real deployments instead deliver small, heavily compressed and partially occluded faces to hardware with a few-watt power budget. We address this gap in three ways. First, DFEW-D and FERV39k-D form a reproducible degradation protocol that jointly sweeps face resolution, codec constant rate factor and occlusion over five severity levels and three codecs. Second, CoDA-Net is a 6.8 M-parameter network that reads motion vectors and residuals from the coded bitstream, combines them with appearance features from sparse intra frames and aggregates the sequence with a bidirectional state-space core, without reconstructing inter-coded frames. Third, quality-conditioned distillation uses a small quality head to estimate per-frame degradation and scales the supervision from a frozen teacher accordingly. We evaluate on DFEW, FERV39k and MAFW. On DFEW-D, CoDA-Net retains 84.2% of its reference-encoding weighted average recall at the most severe degradation level, against 66.7% for the strongest baseline, and runs at 32.3 clips/s on a Jetson Orin Nano and 5.4 clips/s on a Raspberry Pi 5. This suffices for stride-one prediction on the Jetson and for non-overlapping windows on the Pi 5. CoDA-Net is thus a robustness–efficiency trade-off for edge hardware, not a replacement for server-scale models.
People on the autism spectrum and other neurodivergent people may experience sensory-rich environments as distressing, while teachers and care professionals must often identify and respond to environmental discomfort with limited objective information. This study reports the human-centered requirement elicitation, prototype development, and quantitative pilot evaluation of an environmental-monitoring system that relies on rule-based thresholds. Across three case-study settings, 125 stakeholder participations informed the development and design of a smart sensor system measuring sound, light, temperature, and air quality. The subsequent pre–post pilot study was conducted at four Austrian and Italian centers from February to June 2026. In total, 42 clients or pupils and 26 teachers or care professionals participated in the evaluation measures of the pilot study. Post-study acceptance was positive: overall meCUE user experience scores ranged from 4.88 to 4.94 on a seven-point scale, and AttrakDiff overall attractiveness averaged 1.21 (SD = 1.06) on a scale from −3 to +3. The results support the usability and perceived value of the prototype to maintain supportive environmental conditions.
