Search Results (50)

Foundation models, exemplified by the Segment Anything Model (SAM), have revolutionized object segmentation with their impressive zero-shot capabilities. The recent SAM2 extended these abilities to the video domain, utilizing an object pointer and memory attention to maintain temporal segment consistency. However, a critical limitation of SAM2 is its vulnerability to error accumulation, where an initial incorrect mask can propagate through subsequent frames, leading to tracking failure. To address this, we propose a novel method that actively monitors the temporal segment consistency of masks by evaluating the distance of object pointers across frames. When a potential error is detected via a sharp increase in distance, our method triggers a particle filter based re-inference module. This framework models object’s motion to predict a corrected bounding box, effectively guiding the model to recover the valid mask and preventing error propagation. Extensive zero-shot evaluations on DAVIS, LVOS v2, YouTube-VOS and qualitative results show that the proposed, parameter-free procedure consistently improves temporal coherence, raising mean IoU by 0.1 on DAVIS, by 0.13 on the LVOS v2 train split and 0.05 on the LVOS v2 validation split, and by 0.02 on YouTube-VOS, thereby offering a simple and effective route to more robust video object segmentation with SAM2. Full article

Post-exertional malaise (PEM) is a defining symptom of Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS), yet its molecular underpinnings remain elusive. This study investigated the temporal–longitudinal DNA methylation changes associated with PEM using a structured two-day maximum repeated effort cardiopulmonary exercise testing (CPET) protocol involving pre- and two post-exercise blood samplings from five ME/CFS patients. Cardiopulmonary measurements revealed complex heterogeneous profiles among the patients compared to typical healthy controls, and VO₂ peak indicated all patients had poor normative fitness. The switch to anaerobic metabolism occurred at a lower workload in some patients on Day Two of the test. Reduced Representation Bisulphite Sequencing followed by analysis with Differential Methylation Analysis Package-version 2 (DMAP2) identified differentially methylated fragments (DMFs) present in the DNA genomes of all five ME/CFS patients through the exercise test compared with ‘before exercise’. With further filtering for >10% methylation differences, there were early DMFs (0–24 h after first exercise test) and late DMFs between (24–48 h after the second exercise test), as well as DMFs that changed gradually (between 0 and 48 h). Of these, 98% were ME/CFS-specific, compared with the two healthy controls accompanying the longitudinal study. Principal component analysis illustrated the three distinct clusters at the 0 h, 24 h, and 48 h timepoints, but with heterogeneity among the patients within the clusters, highlighting dynamic methylation responses to exertion in individual patients. There were 24 ME/CFS-specific DMFs at gene promoter fragments that revealed distinct patterns of temporal methylation across the timepoints. Functional enrichment of ME-specific DMFs revealed pathways involved in endothelial function, morphogenesis, inflammation, and immune regulation. These findings uncovered temporally dynamic epigenetic changes in stress/immune functions in ME/CFS during PEM and suggest molecular signatures with potential for diagnosis and of mechanistic significance. Full article

This paper focuses on the switching and modulation techniques of terahertz waves, develops ${\mathrm{VO}}_2$ thin-film materials with an SPP structure, and uses terahertz time-domain spectroscopy (THz-TDS) to study the semiconductor–metal phase transition characteristics of ${\mathrm{VO}}_2$ thin films, especially the photoinduced semiconductor–metal phase transition characteristics of silicon-based ${\mathrm{VO}}_2$ thin films. The optical modulation characteristics of silicon-based ${\mathrm{VO}}_2$ thin films to terahertz waves under different light excitation modes, such as continuous light irradiation at different wavelengths and femtosecond pulsed laser irradiation, were analyzed. Combining the optical modulation characteristics of silicon-based ${\mathrm{VO}}_2$ thin films with the filtering characteristics of SPP structures, composite structures of ${\mathrm{VO}}_2$ thin films with metal hole arrays, composite structures of ${\mathrm{VO}}_2$ thin films with metal block arrays, and silicon-based ${\mathrm{VO}}_2$ microstructure arrays were designed. The characteristics of this dual-function device were tested experimentally. The experiment proves that the ${\mathrm{VO}}_2$ film material with an SPP structure has a transmission rate dropping sharply from 32% to 1% under light excitation; the resistivity changes by more than six orders of magnitude, and the modulation effect is remarkable. By applying the SPP structure to the ${\mathrm{VO}}_2$ material, the material can simultaneously possess modulation and filtering functions, enhancing its optical performance in the terahertz band. Full article

The Segment Anything Model 2 (SAM2) has achieved state-of-the-art performance in pixel-level object segmentation for both static and dynamic visual content. Its streaming memory architecture maintains spatial context across video sequences, yet struggles with long-term tracking due to its static inference framework. SAM 2’s fixed temporal window approach indiscriminately retains historical frames, failing to account for frame quality or dynamic motion patterns. This leads to error propagation and tracking instability in challenging scenarios involving fast-moving objects, partial occlusions, or crowded environments. To overcome these limitations, this paper proposes SAM2Plus, a zero-shot enhancement framework that integrates Kalman filter prediction, dynamic quality thresholds, and adaptive memory management. The Kalman filter models object motion using physical constraints to predict trajectories and dynamically refine segmentation states, mitigating positional drift during occlusions or velocity changes. Dynamic thresholds, combined with multi-criteria evaluation metrics (e.g., motion coherence, appearance consistency), prioritize high-quality frames while adaptively balancing confidence scores and temporal smoothness. This reduces ambiguities among similar objects in complex scenes. SAM2Plus further employs an optimized memory system that prunes outdated or low-confidence entries and retains temporally coherent context, ensuring constant computational resources even for infinitely long videos. Extensive experiments on two video object segmentation (VOS) benchmarks demonstrate SAM2Plus’s superiority over SAM 2. It achieves an average improvement of 1.0 in J&F metrics across all 24 direct comparisons, with gains exceeding 2.3 points on SA-V and LVOS datasets for long-term tracking. The method delivers real-time performance and strong generalization without fine-tuning or additional parameters, effectively addressing occlusion recovery and viewpoint changes. By unifying motion-aware physics-based prediction with spatial segmentation, SAM2Plus bridges the gap between static and dynamic reasoning, offering a scalable solution for real-world applications such as autonomous driving and surveillance systems. Full article

In this paper, we propose a multi-mode switchable ultra-wideband terahertz absorber based on patterned graphene and VO₂ by designing a graphene pattern composed of a large rectangle rotated 45° in the center and four identical small rectangles in the periphery, as well as a VO₂ layer pattern composed of four identical rectangular boxes and small rectangles embedded in the dielectric layer. VO₂ can regulate conductivity via temperature, the Fermi level of graphene depends on the external voltage, and the graphene layer and VO₂ layer produce resonance responses at different frequencies, resulting in high absorption. The proposed absorption microdevices have three modes: Mode 1 (2.52–4.52 THz), Mode 2 (3.91–9.66 THz), and Mode 3 (2.14–10 THz), which are low-band absorption, high-band absorption, and ultra-wideband absorption. At 2.96 THz in Mode 1, the absorption rate reaches 99.98%; at 8.04 THz in Mode 2, the absorption rate reaches 99.76%; at 5.04 THz in Mode 3, the absorption rate reaches 99.85%; and at 8.4 THz, the absorption rate reaches 99.76%. We explain the absorption mechanism by analyzing the electric field distribution and local plasma resonance, and reveal the high-performance absorption mechanism by using the relative impedance theory. In addition, absorption microdevices have the advantages of polarization insensitivity, incident angle insensitivity, multi-mode switching, ultra-wideband absorption, large manufacturing tolerance, etc., and have potential research and application value in electromagnetic stealth devices, filters and optical switches. Full article

A dual-band tunable terahertz electromagnetically induced transparency (EIT) metamaterial is introduced. The EIT metamaterial consists of two rectangular split rings, two metal strips, and a patterned vanadium dioxide (VO₂) located at the back. The rectangular split rings serve as the bright resonator to generate two resonance valleys at distinct frequencies. The metal strips act as the dark resonator and are indirectly activated via the coupling influence of the bright resonator. The EIT metamaterial’s response mechanism is analyzed via the field effect and the two-particle model, with theoretical fitting results showing strong agreement with the simulation results. Moreover, VO₂’s conductivity is altered to dynamically control the EIT effect in both frequency bands. Two transparency windows, with modulation depths of 70% and 75%, are observed as the conductivity of VO₂ decreases. Simultaneously, the simulation results reveal a favorable slow light effect, with group delays reaching 51 ps and 74 ps at the transparency windows. The proposed metamaterial holds considerable promise for future modulator, filter, and slow light device applications. Full article

Global warming has driven a marked increase in forest fire occurrences, underscoring the critical need for timely and accurate detection to mitigate fire-related losses. Existing forest fire detection algorithms face limitations in capturing flame and smoke features in complex natural environments, coupled with high computational complexity and inadequate lightweight design for practical deployment. To address these challenges, this paper proposes an enhanced forest fire detection model, YOLOv8n-SMMP (SlimNeck–MCA–MPDIoU–Pruned), based on the YOLO framework. Key innovations include the following: introducing the SlimNeck solution to streamline the neck network by replacing conventional convolutions with Group Shuffling Convolution (GSConv) and substituting the Cross-convolution with 2 filters (C2f) module with the lightweight VoV-based Group Shuffling Cross-Stage Partial Network (VoV-GSCSP) feature extraction module; integrating the Multi-dimensional Collaborative Attention (MCA) mechanism between the neck and head networks to enhance focus on fire-related regions; adopting the Minimum Point Distance Intersection over Union (MPDIoU) loss function to optimize bounding box regression during training; and implementing selective channel pruning tailored to the modified network architecture. The experimental results reveal that, relative to the baseline model, the optimized lightweight model achieves a 3.3% enhancement in detection accuracy (mAP@0.5), slashes the parameter count by 31%, and reduces computational overhead by 33%. These advancements underscore the model’s superior performance in real-time forest fire detection, outperforming other mainstream lightweight YOLO models in both accuracy and efficiency. Full article

Energy-selective surfaces (ESSs) have gained attention as an advanced electromagnetic protection technology. This review discusses the evolution of ESSs, focusing on four key areas: frequency bandwidth expansion, material innovations, functional enhancements, and application diversification. ESSs have evolved from narrowband designs to providing ultra-wideband protection, covering L-band to K-band frequencies. New designs, including non-reciprocal mechanisms and cascaded filters, enhance the shielding efficiency. Material advancements like the use of vanadium dioxide (VO₂) and micro–nano fabrication techniques have reduced costs and improved performance, enabling higher-frequency applications. Future developments aim to overcome the current limitations, offering a broader bandwidth, higher power tolerance, and faster response times. ESSs play a key role in integrated electromagnetic protection systems. Full article

This paper proposes a switchable and tunable terahertz metamaterial absorber utilizing a graphene-VO₂ layered structure. The design employs reconfigurable seven-layer architecture from top to bottom as (topaz/VO₂/topaz/Si/graphene/topaz/Au). CST software 2018 was used to simulate the absorption properties of terahertz waves (0–14 THz). The proposed metamaterial exhibits dual functionalities depending on the VO₂ phase state. In the insulating state, the design achieves a tri-band response with distinct peaks at 3.12 THz, 5.65 THz, and 7.24 THz. Conversely, the VO₂’s conducting state enables ultra-broadband absorption from 2.52 THz to 11.62 THz. Extensive simulations were conducted to demonstrate the tunability of absorption: Simulated absorption spectra were obtained for broadband and multi-band states. Electric field distributions were analyzed at resonance frequencies for both conducting and insulating states. The impact was studied of VO₂ conductivity, loss tangent, and graphene’s chemical potential on absorption. The influence was investigated of topaz layer thickness on the absorption spectrum. Absorption behavior was examined of VO₂ under different states and layer configurations. Variations were analyzed of absorption spectra with frequency, polarization angle, and incident angle. The proposed design used for the detection of cervical and breast cancer detection and the sensitivity is about is 0.2489 THz/RIU. The proposed design holds significant promise for real-world applications due to its reconfigurability. This tunability allows for tailoring absorption properties across a broad terahertz range, making it suitable for advanced devices like filters, modulators, and perfect absorbers. Full article

Temperature sensors play important roles in wide-spreading human activities. The non-contact method of using temperature sensors offers significant advantages but faces challenges in detection precision. In this work, a double-layer asymmetric terahertz (THz) metamaterial combined with phase transition oxide was proposed to realize non-contact temperature sensor with high sensitivity. The metamaterial exhibited band-stop filtering effects in the simulated transmission spectra. Temperature changes induced a reversible phase transition in VO₂, resulting in altered conductivity. The numerical results indicated that the S21 parameter increases from −44.33 dB to −4.78 dB at a frequency of 1.22 THz as the conductivity of the VO₂ film increases from 10 to 5000 S/m, achieving a modulation depth of 89%. In addition, the 86 nm thick VO₂ film underwent a phase transition in the temperature range of 54.93 °C to 66.93 °C, achieving a sensitivity of 1.82 dB/°C for temperature sensing. This work provided great insights into the development of metamaterials based on high-precision temperature measurement. Full article

Over the past decade, there has been a growing recognition of the importance of indoor environmental quality (IEQ) in influencing occupant productivity. Researchers have studied various buildings, including offices, schools, hospitals, and residential settings, to understand the relationship between IEQ and productivity outcomes. Studies have taken a multifactorial approach, considering multiple aspects of IEQ. Evidence from the literature review suggests that the quality of the indoor environment is an essential factor that affects the productivity of building occupants, and it is one of the fundamental issues in the development of societies. This area of research requires the responsible participation of researchers at all levels, as there is significant scope to contribute to knowledge. Therefore, this study aims to conduct a bibliometric analysis of the published literature on indoor environmental quality and its impact on building occupant productivity through the scientific literature available from one of the largest and most famous academic databases, Scopus; the study was determined in 2011 to 2023. The search used differential thresholds for IEQ keywords affecting building occupant productivity. Three discrete queries were performed, resulting in approximately 3861 publications. These were filtered by reducing false positives and excluding publications irrelevant to the research topic. The final results were 72 publications. This study also used Excel and VOS viewer to analyse and create graphs and network visualisation maps to show the growth of publications and their types, active countries and institutions for recovered publications, international collaboration, author keywords, active journals, and citation analysis. This study can significantly advance our understanding of building occupant productivity and enhance quality of life and work. Evaluating the research outputs is essential for highlighting contributions to knowledge and global collaboration in this research area. The potential impact of this study is not just theoretical. It can shape the future of our built environments and the lives of those occupying them. Full article

Referring video object segmentation (R-VOS) is a fundamental vision-language task which aims to segment the target referred by language expression in all video frames. Existing query-based R-VOS methods have conducted in-depth exploration of the interaction and alignment between visual and linguistic features but fail to transfer the information of the two modalities to the query vector with balanced intensities. Furthermore, most of the traditional approaches suffer from severe information loss in the process of multi-scale feature fusion, resulting in inaccurate segmentation. In this paper, we propose DCT, an end-to-end decoupled cross-modal transformer for referring video object segmentation, to better utilize multi-modal and multi-scale information. Specifically, we first design a Language-Guided Visual Enhancement Module (LGVE) to transmit discriminative linguistic information to visual features of all levels, performing an initial filtering of irrelevant background regions. Then, we propose a decoupled transformer decoder, using a set of object queries to gather entity-related information from both visual and linguistic features independently, mitigating the attention bias caused by feature size differences. Finally, the Cross-layer Feature Pyramid Network (CFPN) is introduced to preserve more visual details by establishing direct cross-layer communication. Extensive experiments have been carried out on A2D-Sentences, JHMDB-Sentences and Ref-Youtube-VOS. The results show that DCT achieves competitive segmentation accuracy compared with the state-of-the-art methods. Full article

This study proposes a dual-functional terahertz device based on the Dirac semimetal, serving as both a sensing element and a band-pass filter. The device’s operating mode can switch between these two functions by utilizing the phase transition property of vanadium dioxide (VO₂). When VO₂ is in the insulating state, the device functions as a sensing element. The simulation results demonstrate an impressive refractive index sensitivity of 374.40 GHz/RIU (Refractive Index Unit). When VO₂ is in the metallic state, the device functions as a band-pass filter, exhibiting a center frequency of 2.01 THz and a 3 dB fractional bandwidth of 0.91 THz. The integration of these dual functionalities within a single terahertz device enhances its utility in both sensing and filtering applications. Full article

Scientific reports have diversified enormously in the various thematic areas of the scientific world, generating errors and biases in the searches and directing the various investigations. For this reason, this article provides an approach, from the point of view of bibliometric analysis, to reveal the trends of agricultural waste for its potential use as fuel in generating electrical energy. This research has used the programs RStudio, VosViewer, and Excel for the compilation and analysis of data, whose data were extracted from Scopus during the period from 2013 to 15 March 2024, filtering all types of publications that were not original articles and English language. Scientific reports have found that biodigesters are constantly evolving, improving valves, studying and analyzing different types of agricultural waste, and using microorganisms to accelerate the fermentation process of agricultural waste. The terms biogas (330 occurrences), anaerobic digestion (214 occurrences), and agriculture (212 occurrences) were the words with the highest occurrences. At the same time, the author that stood out was Liu Y., who had an H index of 6 and 117 citations in his six articles published in the Scopus database. Moreover, China (1900 citations) was the country with the highest citation numbers, followed by the United States (1060 citations) and India (967 citations). The designs of biogas production increase efficiency and can increase biodigesters’ performance. The research also reveals the different types of development and trends that stood out and emerged in the last decade, such as the authors who have had the most impact on this topic that has recently emerged and the countries that have obtained the most significant number of publications on the topic. Full article

Tracking and segmenting small targets in remote sensing videos on edge devices carries significant engineering implications. However, many semi-supervised video object segmentation (S-VOS) methods heavily rely on extensive video random-access memory (VRAM) resources, making deployment on edge devices challenging. Our goal is to develop an edge-deployable S-VOS method that can achieve high-precision tracking and segmentation by selecting a bounding box for the target object. First, a tracker is introduced to pinpoint the position of the tracked object in different frames, thereby eliminating the need to save the results of the split as other S-VOS methods do, thus avoiding an increase in VRAM usage. Second, we use two key lightweight components, correlation filters (CFs) and the Mobile Segment Anything Model (MobileSAM), to ensure the inference speed of our model. Third, a mask diffusion module is proposed that improves the accuracy and robustness of segmentation without increasing VRAM usage. We use our self-built dataset containing airplanes and vehicles to evaluate our method. The results show that on the GTX 1080 Ti, our model achieves a J&F score of 66.4% under the condition that the VRAM usage is less than 500 MB, while maintaining a processing speed of 12 frames per second (FPS). The model we propose exhibits good performance in tracking and segmenting small targets on edge devices, providing a solution for fields such as aircraft monitoring and vehicle tracking that require executing S-VOS tasks on edge devices. Full article