Search Results (1,032)

In response to issues such as disordered airflow distribution and prominent local hotspots in campus non-standard data centers, this study took a non-standard raised-floor air-supply data center at a university in Hangzhou as the research object, and used a combination of on-site measurements and computational fluid dynamics (CFD) numerical simulation to investigate the optimization of the thermal environment. The temperature and air velocity of the data center were measured using a handheld hot-wire anemometer, and a standard k-ε turbulence model was established on the 6SigmaDC platform (now Cadence Reality DC Design Pro, version 2024.1). Model accuracy was confirmed through grid independence verification with three mesh levels and statistical error metrics (MAE, MBE, RMSE) across multiple measurement zones. The results show that the mean absolute error of temperature does not exceed 0.9 °C in all zones and the mean absolute error of air velocity does not exceed 0.20 m/s, indicating that the model effectively reproduces the airflow distribution and thermal environment of the data center. On this basis, to address the uneven airflow distribution in the underfloor plenum, an optimization strategy was proposed that involved the installation of composite baffles and the coordinated adjustment of variable floor tile openings. Eight representative simulation scenarios were designed, with the coefficient of variation and air supply uniformity index as evaluation indicators. Results indicate that the combined effect of perforated baffles and variable floor tile openings is the optimal strategy, reducing the range of net airflow among air supply outlets from 0.100 to 0.077 m³/s, decreasing the coefficient of variation from 12.8% to 10.8%, and increasing the air supply uniformity index by 10.7%. Whole-room thermal environment verification shows that the optimal scheme reduces the supply heat index (SHI) from 0.42 to 0.35, with an estimated PUE reduction of about 0.03, achieving both airflow uniformity improvement and energy-saving benefits. By improving the cooling efficiency and reducing the PUE, this retrofit strategy contributes to the sustainable operation of small-to-medium-sized campus data centers, supporting energy efficiency and carbon footprint reduction goals under green campus and low-carbon initiatives. Full article

Background: Minimally invasive cardiac valve surgery has emerged as a preferred approach in selected patients, yet optimal postoperative extubation timing remains debated. This systematic review and meta-analysis examined clinical outcomes associated with extubation in the operating room (OR) versus the intensive care unit (ICU) among adult patients undergoing minimally invasive cardiac valve surgery. Methods: The study was conducted according to PRISMA guidelines. A single unit of analysis was applied throughout. Pooled odds ratios were computed with the Mantel–Haenszel random-effects method; where a study reported only a matched or covariate-adjusted estimate, that estimate was reserved for a prespecified sensitivity analysis using the generic inverse-variance method. Results: Five observational studies (2023–2025) including 1101 OR-extubated and 899 ICU-extubated patients from high-volume centers with fast-track or enhanced recovery pathways were included. OR extubation was associated with lower odds of reintubation (OR 0.40; 95% CI 0.24–0.69; I² = 0%), postoperative delirium (OR 0.47; 95% CI 0.31–0.72; I² = 0%), and pneumonia (OR 0.30; 95% CI 0.16–0.53; I² = 0%). No significant differences were observed for new-onset atrial fibrillation, stroke, or reoperation for bleeding. Thirty-day mortality was reported by four of the five studies and comprised few events (5 of 1043 ORE versus 17 of 645 ICE across the four studies reporting this outcome); given the small number of events, the concentration of deaths in the higher-risk ICU-extubated patients, and the reliance of the pooled estimate on two confounded cohorts, this difference is not interpretable as a treatment effect, and no pooled odds ratio is reported here. Length of stay was consistently shorter after OR extubation but was not pooled because of extreme heterogeneity (I² = 96–100%). Sensitivity analyses using adjusted estimates attenuated the associations for reintubation and pneumonia, consistent with substantial confounding by indication. Conclusions: In appropriately selected patients undergoing minimally invasive valve surgery, OR extubation is feasible and is associated with a recovery profile at least comparable to that of ICU extubation. Because extubation location was determined largely by intraoperative and early postoperative stability, these associations should be read as reflecting patient selection rather than a causal benefit of the strategy. The findings support the feasibility of OR extubation in appropriately selected patients at experienced centers and motivate prospective, ideally randomized, evaluation. Full article

Coffee leaf diseases impair photosynthesis, thereby degrading the chemical composition and flavor quality of coffee beans. However, these resulting quality defects are often not visible from the appearance of green beans alone, compelling e-commerce quality control to trace back to leaf disease detection at the production origin. This focus on “quality traceability” imposes requirements on computer vision technologies that fundamentally differ from those of conventional pesticide-application-oriented detection: prioritizing high precision over high recall, replacing simple classification with severity grading, and necessitating the integration of variety and origin metadata. This paper conducts a systematic review of 53 relevant studies published from January 2020 to March 2026, examining existing data resources, model architectures, and industrial adaptability through the lens of e-commerce quality traceability. Our review highlights three major findings: (1) existing datasets could be further enriched in variety labeling, severity scoring, and origin metadata; (2) current models, predominantly focused on classification, have room for closer alignment with traceability requirements regarding optimization objectives, task definitions, and output formats; and (3) cutting-edge technologies, including semantic segmentation, multimodal fusion, and visual foundation models, offer viable pathways to bridge these gaps. This review provides standardized technical evaluation criteria and clear optimization directions for origin inspection, batch grading and whole-chain traceability management of coffee agricultural e-commerce platforms. Full article

Object detection remains a fundamental challenge in computer vision and plays a pivotal role in safety-critical medical applications, including surgical instrument recognition and operating-room workflow automation. This study presents a comprehensive comparative evaluation of five attention mechanisms—Squeeze-and-Excitation (SE), Convolutional Block Attention Module (CBAM), Efficient Channel Attention (ECA), Simple Attention Module (SimAM), and an enhanced multi-kernel Spatial Pyramid Pooling Fast module (SPPF+)—integrated into the YOLOv26n backbone, together with two neck-level attention variants (ECA-Neck and CBAM-Neck). A total of 16 model configurations were systematically investigated on a 22-class dental surgical instrument detection dataset under both default training settings and hyperparameter configurations optimized using the Optuna Tree-structured Parzen Estimator (TPE), enabling a rigorous full-factorial ablation study. Experimental results demonstrate that TPE-based hyperparameter optimization consistently enhances detection performance across all architectures. Among the evaluated models, CBAM-Opt achieved the highest detection accuracy, attaining an mAP@50 of 0.959 and an F1-score of 0.913, although the margins among the top optimized configurations fall within run-to-run variability. In contrast, Base-Opt delivered the strongest strict-localization capability with an mAP@50–95 of 0.800, highlighting the competitive performance of the baseline architecture when appropriately optimized. Notably, the parameter-free SimAM module exhibited the largest improvement following optimization ($\Delta$mAP@50 = +0.040), indicating a pronounced sensitivity to training configuration. Furthermore, neck-level attention integration achieved performance comparable to backbone-based attention, with ECA-Neck-Opt reaching an mAP@50 of 0.959, suggesting an effective alternative that preserves pretrained feature representations while maintaining high detection accuracy. Beyond performance evaluation, this work provides a unified benchmarking framework for attention mechanisms in medical object detection, accompanied by computational complexity analysis and practical architectural insights. The findings establish evidence-based guidelines for selecting attention modules in resource-aware surgical vision systems and contribute toward the development of more accurate and reliable computer-assisted clinical workflows. Full article

Property condition assessment is a critical step in residential real estate investment underwriting, motivating after-repair value (ARV) estimates and rehabilitation cost projections. Traditional approaches rely on in-person inspections or manual photo review by experienced investors—processes that are time-consuming, subjective, and do not scale. Prior computer vision work on building analysis has focused on structural defect detection using convolutional neural networks but has not addressed the holistic, room-level condition assessment needed for residential investment decision-making. This paper presents a per-room analysis framework that leverages multimodal large language models (MLLMs) to assess the condition of residential properties from photographs. The framework analyzes each photo independently at the room level—detecting the room type, condition category, condition score, material features, and visible issues. Condition output is intended to feed a separate downstream rehabilitation cost and ARV estimation model that is outside the scope of this paper; the present empirical evaluation is restricted to per-photo condition assessment and inter-rater agreement with human experts. I evaluate the framework on two complementary datasets: (i) a primary per-image condition evaluation on 57 photographs from 14 real off-market properties in the Memphis, TN MSA, spanning three condition tiers (Fixer, Outdated, Standard), with independent labels from two experienced real estate investors; (ii) a secondary room classification evaluation on the public REI Dataset (51 attempted, 39 successful, 12 HTTP-503 failures). The room classification accuracy was 76.5% intention-to-analyze on REI (100% per-protocol on the 39 successful calls; 23.5% API failure rate) and 82.5% on the concierge dataset. The inter-rater agreement on the concierge dataset, with 95% bootstrap CIs (5000 resamples) and Spearman’s $\rho$ as primary score statistic, was as follows: Cohen’s $\kappa =0.773$ ($95\%$ CI $[0.64,0.90]$) between Labeler A and the MLLM (weighted $\kappa =0.853$ $[0.76,0.94]$; $\rho =0.906$); and $\kappa =0.502$ $[0.35,0.66]$ between Labeler B and the MLLM ($\rho =0.858$); both bracket the human–human reliability of $\kappa =0.590$ $[0.42,0.74]$ ($\rho =0.807$). The MLLM’s $\kappa$ asymmetry across the two labelers is statistically significant ($\Delta \kappa =0.271$, $95\%$ bootstrap CI $[0.115,0.429]$, $p=0.0004$), which I attribute to plausible training distribution and labeling style differences. A blind re-labeling sensitivity analysis on a stratified 15-image subsample yields anchoring-corrected $\kappa$ estimates of approximately $0.65$ (Labeler A) and $0.35$ (Labeler B); the headline anchored values therefore sit at the upper bound of plausible blind-equivalent agreement. Failure modes concentrate at the Outdated tier and at the Outdated–Standard boundary, where humans themselves disagree most, indicating intrinsic taxonomy ambiguity rather than a model artifact. I make no claim to multi-market generalization and present multi-market extension as ongoing work. Full article

Ferroelectric field-effect transistors, which rely on ferroelectric polarization reversal to modulate the channel resistance, hold great promise for nonvolatile memory and neuromorphic computing. The polarization switching dynamics are critical for achieving high-speed, high-bit-density neuromorphic hardware. Here, we report a 1D-2D asymmetric heterojunction composed of a single-element tellurium (Te) nanowire and a magnetic metal, Fe₃GaTe₂. Piezoresponse force microscopy reveals reversible polarization switching at room temperature. Utilizing this ferroelectric heterojunction, we construct ferroelectric semiconductor field-effect transistors that exhibit tunable resistance states exceeding 7 bits, featuring an on/off ratio of 10³, a retention time exceeding 10³ s, and ultrafast switching down to 20 ns. Moreover, the transistor enables accurate recognition of six kinds of micro-defects with an accuracy of 97.1% on the workpiece surface by convolutional neural network. This work establishes the intrinsic relationship between ferroelectric polarization and resistance modulation, providing a device platform for next-generation multilevel storage and ultrafast neuromorphic computing networks. Full article

Background: Environmental hygiene in operating rooms is essential for infection prevention, but visual inspection alone may not reliably identify missed cleaning. Aim: To assess the feasibility of fluorescent gel marker monitoring with ultraviolet light, combined with a nurse-led educational and feedback intervention, on the cleaning of high-touch operating room surfaces. Methods: A quasi-experimental, nonrandomized, single-group pre-post study was conducted in five operating rooms of a district general hospital in southern Portugal. Forty-two cleaning episodes and 420 surface assessments were evaluated across three sequential phases: blinded baseline monitoring, post-training monitoring after a 45-min nurse-led workshop, and feedback-based monitoring using an ultraviolet walk-through. The primary outcome was the proportion of surface assessments classified as completely clean. Exploratory Mann–Whitney U tests compared ordinal cleaning outcomes between phases; p-values were nominal and not adjusted for multiple comparisons. Results: Complete cleanliness increased from 44.7% at baseline to 69.3% after training and 89.2% during feedback-based monitoring. The largest observed increases occurred among surfaces with poor baseline performance, particularly anesthesia machine rebreathing bags, IV stands, computer keyboards, and computer mice. Visual inspection identified dust on keyboards and occasional medication residue but missed several surfaces detected by fluorescent marker assessment. Conclusions: Fluorescent marker monitoring with ultraviolet feedback was feasible within the study setting and was accompanied by progressive improvements in the proportion of surfaces classified as completely clean. However, the absence of a concurrent control group precludes attributing these changes exclusively to the intervention. The approach may support nurse-led quality improvement in operating room environmental hygiene. Full article

Standalone use of EAHEs for room cooling is a passive and nature-based alternative to air conditioning technology that can be used to mitigate the increase in electricity and GWP-refrigerant consumption associated with cooling in buildings. EAHEs replacing air conditioning is documented in the technical and research literature. However, for office-room cooling, EAHEs are mostly employed as a support to air conditioning systems for precooling outdoor air. The larger cooling loads and the stricter design conditions commonly used in the sizing of office rooms prevent the most commonly investigated EAHE typologies from operating effectively in standalone cooling mode. To assess the feasibility of alternative typologies, such as large-diameter EAHEs, tools that are capable of modeling the complexity of the coupled heat and moisture transfer between air and soil are particularly valuable. For detailed assessments, researchers typically turn to advanced commercial tools; however, developments in free and open-source scientific programming languages that combine symbolic computation packages with efficient numerical solvers of partial differential equations allow analyses at reduced cost that are comparable to those from commercial tools. This paper shows how one such programming language can be used to study the coupled heat and moisture transfer problem in EAHEs. Starting from the symbolic form of the mathematical problem, the numerical implementation is described and validated with monitoring data from an existing large-diameter EAHE. Using the validated computational model, the paper proceeds to study the sensitivity of load removal in EAHEs operating in standalone and precooling cooling modes, highlighting fundamental differences between both operating modes, identifying the most relevant design parameters and providing guidance on the conditions under which an EAHE enables self-sufficient cooling of office buildings. The results show how, for a hot and dry climate, standalone EAHEs with large diameters (∼1 m), buried at depths larger than 3 m, allow the removal of up to 20 kWh/m² of room sensible cooling loads, a level that is consistent with the cooling demand of low-energy office buildings. Full article

Exhibition layouts in museums strongly influence visitors’ learning and viewing experiences, yet their evaluation often relies on curators’ subjective judgments. This study aims to provide objective behavioral data through visitor flow analysis to support evidence-based exhibition design. Single-board computers installed in the museum exhibition rooms serve as sensors to detect humans in captured images using machine learning. Detected positions are projected onto a floor map using a homography transformation to reconstruct visitors’ spatiotemporal behavior. The proposed system enables server-side real-time analysis and visualization, allowing curators to quickly and objectively understand visitor movement patterns. Full article

Green low-carbon data centers operate as coupled cyber-energy systems whose dispatch must coordinate renewable generation, grid exchange, battery storage, cooling load, flexible computing workload, carbon-intensity signals, and reliability constraints. This study develops and evaluates a safety-constrained deep reinforcement learning framework for source–load–storage coordinated operation of a grid-connected green data center. The operating problem is formulated as a constrained Markov decision process with state variables describing the IT load, deferrable workload backlog, renewable availability, electricity price, marginal carbon intensity, battery state of charge, server-room temperature, reserve margin, and calendar context. The action space covers grid import and export, renewable utilization, storage charge and discharge, workload shifting, and cooling control. The learning architecture combines a constrained actor–critic policy, adaptive Lagrangian safety critics, and a control barrier function (CBF)-based action shield that projects unsafe actions onto an explicitly defined operating set before plant execution. The shield is specified as a low-dimensional quadratic projection over state-dependent SOC, thermal, reserve, SLA, and grid-interface constraints, while cumulative risks are priced through Lagrangian safety budgets during policy training. The evaluation uses a controlled and auditable benchmark simulation with normalized public-data-compatible profiles, declared scenarios, random seeds, neural-network settings, and mechanism-matched baselines; it is not a telemetry-based verification or hardware certification of a deployed data center. Within this declared benchmark, the proposed safe DRL controller produces a simulated 13.1% emission reduction relative to the Rule-based controller, 95.8% renewable utilization, a normalized annual cost of 0.91, and fewer boundary contacts than the tested unconstrained, Lagrangian-only, and shield-only PPO variants. These percentages are simulator outputs relative to the stated benchmark and must not be interpreted as measured field savings. The results show how separating reward learning, cumulative safety pricing, and one-step engineering projection changes low-carbon dispatch within the specified model. Full article

Indoor wall surface flatness is a mandatory acceptance item in building decoration work, but it is still commonly checked using a 2 m straightedge and a feeler gauge. This manual method samples only a limited number of locations, depends on inspector judgement, and provides no full-surface spatial record. This study proposes a terrestrial laser scanning (TLS)-based automated workflow for wall surface flatness inspection that remains consistent with the Chinese national standard GB 50210-2018. After point-cloud registration and wall segmentation, a least-squares best-fit plane is established for each wall, and the signed point-to-plane deviation field is computed. A virtual 2 m straightedge operator is then applied at multiple positions and orientations to reproduce the code-defined measurement in which the maximum gap under the straightedge is taken as the flatness reading. The method was validated against paired manual measurements on a reference wall, showing a mean difference of 0.00 mm, a standard deviation of 0.20 mm, an RMSE of 0.20 mm, and a correlation coefficient of 0.98, with identical accept/reject decisions at the 4 mm limit. A residential case study involving four inspection zones and 42 measurement points further demonstrated that the workflow can generate code-comparable flatness readings, full-surface deviation maps, and localized rework guidance. The overall pass rate was 92.9%, with all non-conforming points concentrated in the living room, and the on-site inspection time was approximately halved compared to manual checking. The results indicate that the proposed TLS-based virtual-straightedge method provides a practical, traceable, and standard-compliant alternative for automated wall surface flatness acceptance. Full article

Background/Objectives: Ground-glass pulmonary nodules are often nonpalpable and not visible during thoracoscopic surgery, making accurate localization important for achieving adequate pathological margins. However, no clinically validated localization-error threshold has been established. We evaluated the association between Dmn and pathological margin adequacy after image-guided thoracoscopic pulmonary nodule resection and sought to identify a clinically relevant localization-accuracy threshold. Methods: We retrospectively reviewed 169 patients in a predefined peripheral-lesion cohort who underwent cone-beam computed tomography-guided pulmonary nodule localization using hook-wire placement or dye marking in a hybrid operating room, followed by thoracoscopic wedge resection. Dmn was defined as the shortest three-dimensional Euclidean distance from the localization needle tip to the tumor margin. Pathological margin adequacy was defined as a resection margin equal to or greater than the maximum tumor diameter. Logistic regression was used to identify predictors of pathological margin inadequacy, and receiver operating characteristic analysis was performed to determine the optimal Dmn cutoff. Results: Dmn was independently associated with pathological margin inadequacy (odds ratio, 1.617; 95% confidence interval, 1.356–1.928; p < 0.001). Receiver operating characteristic analysis yielded an area under the curve of 0.827 and an optimal Dmn cutoff of 4.90 mm, with a sensitivity of 71.0% and specificity of 94.2%. Pathological margin inadequacy occurred more frequently in patients with Dmn ≥ 4.9 mm than in those with Dmn < 4.9 mm (73.3% vs. 6.5%, p < 0.001). Conclusions: In this single-center cohort, a Dmn of approximately 4.9 mm was associated with pathological margin inadequacy after cone-beam computed tomography-guided pulmonary nodule localization and thoracoscopic wedge resection. This value may serve as a candidate intraoperative warning threshold for margin reassessment; however, prospective external validation is required before routine clinical implementation. Full article

Background/Objectives: Hybrid operating rooms combine advanced intraoperative imaging, endovascular capabilities, and multidisciplinary resources within a single procedural environment. However, their use in otolaryngology remains insufficiently characterized. This study evaluated institutional patterns of hybrid operating room use, principal clinical indications, multidisciplinary involvement, and perioperative resource utilization in otolaryngology. Methods: We conducted a retrospective single-center study of eligible otolaryngologic procedures performed between 1 October 2018 and 31 December 2025. Patient characteristics, operative sites, hybrid operating room applications, multidisciplinary involvement, intraoperative blood loss, postoperative intensive care unit admission, and length of hospital stay were analyzed. Results: A total of 55 unique procedures were included. The median age was 46.0 years (interquartile range, 33.0–58.5 years; range, 5–76 years), and 33 patients (60.0%) were male. Computed tomography-based localization and navigation represented the predominant application, accounting for nearly three-quarters of procedures. The sinonasal cavity, nasopharynx, and skull base were the most frequently treated anatomical regions, comprising approximately 60% of operative sites. Angiography and endovascular intervention constituted the second most common application. Multidisciplinary collaboration, most frequently involving cardiovascular surgery and interventional radiology, was required in nearly one-quarter of procedures. Procedures relying primarily on intraoperative imaging were associated with a median estimated blood loss of 20 mL and a median hospital stay of 3 days. Cases requiring vascular, cardiopulmonary, or other advanced hybrid capabilities showed greater postoperative resource utilization, including more frequent intensive care admission and longer hospitalization. Conclusions: The hybrid operating room served as a versatile platform for image-guided, vascular, and multidisciplinary procedures in otolaryngology. Its capabilities were used during the management of anatomically complex and high-acuity cases, while differences in postoperative resource utilization appeared to reflect procedural complexity and baseline clinical risk. Full article

A huge effort has been directed towards research and development of new measurement systems for maximizing comfort and safety in public buildings by monitoring indoor air quality (IAQ). In fact, according to World Health Organization, exposure to chemical, biological, and physical agents in poorly ventilated spaces can lead to psycho-physical discomfort as well as respiratory and neurological diseases. Recent advances in the Internet of Things (IoT) have paved the ground for the design and implementation of distributed measurement systems with higher sensor density and computational capacity. While these systems provide accurate assessments of individual rooms, they do not account for personal exposure to varying air quality levels over time. In public buildings such as schools, universities, and workplaces, occupants frequently move between rooms according to predefined schedules, resulting in heterogeneous exposure patterns. To address this issue, this paper proposes an innovative IAQ measurement technique for public buildings, shifting the focus from room-based assessment to occupant-centered assessment. Unlike wearable or portable personal monitors, the proposed technique infers occupant location from the institutional timetable and combines it with the fixed sensor infrastructure already installed in the rooms, requiring no additional devices to be worn. Individual conditions are quantified through a new personalized metric that integrates instantaneous air quality, cumulative individual exposure over time, and thermal comfort into a single index that is evaluated against occupant-specific thresholds. The technique is validated using real-world data, demonstrating higher potential to ensure safety and comfort compared to the state of the art. Full article

3D point cloud understanding plays a vital role in remote sensing, robotic perception and intelligent scene analysis. However, real-world point cloud data are often affected by sensing noise, incomplete geometry, occlusion, and irregular sampling, posing significant challenges to reliable geometric representation learning and long-range contextual modeling. Existing methods typically rely on fixed neighborhood aggregation or computationally expensive global interaction mechanisms, leaving considerable room for improvement in terms of robustness and efficiency under complex sensing conditions. To address these challenges, we propose GeoSeqNet, a geometry-aware contextual learning framework for robust 3D point cloud analysis. Specifically, an Enhanced Local Operator (ELO) is introduced to strengthen local geometric representation, while a Geometric Encoding Module (GEM) is employed to preserve spatial geometric information during long-range feature interactions. In addition, an Adaptive Gate Fusion (AGF) module is designed to effectively integrate Gate-Scaled LSTM and GRU branches, enabling efficient long-range contextual modeling. By jointly exploiting local geometric cues and long-range contextual information, GeoSeqNet achieves robust feature learning with low computational overhead. Extensive experiments on ModelNet40, ScanObjectNN, and ShapeNetPart demonstrate the effectiveness of GeoSeqNet. The proposed method achieves competitive performance while maintaining a favorable efficiency–accuracy trade-off and exhibits strong robustness in complex real-world scenarios. These results indicate that GeoSeqNet provides an effective and reliable solution for point cloud understanding in challenging sensing environments. Full article