Search

Abstract Skeleton‐based action recognition has received much attention and achieved remarkable achievements in the field of human action recognition. In time series action prediction for different scales, existing methods mainly focus on attention mechanisms to enhance modelling capabilities in spatial dimensions. However, this approach strongly depends on the local information of a single input feature and fails to facilitate the flow of information between channels. To address these issues, the authors propose a novel Temporal Channel Reconfiguration Multi‐Graph Convolution Network (TRMGCN). In the temporal convolution part, the authors designed a module called Temporal Channel Fusion with Guidance (TCFG) to capture important temporal information within channels at different scales and avoid ignoring cross‐spatio‐temporal dependencies among joints. In the graph convolution part, the authors propose Top‐Down Attention Multi‐graph Independent Convolution (TD‐MIG), which uses multi‐graph independent convolution to learn the topological graph feature for different length time series. Top‐down attention is introduced for spatial and channel modulation to facilitate information flow in channels that do not establish topological relationships. Experimental results on the large‐scale datasets NTU‐RGB + D60 and 120, as well as UAV‐Human, demonstrate that TRMGCN exhibits advanced performance and capabilities. Furthermore, experiments on the smaller dataset NW‐UCLA have indicated that the authors’ model possesses strong generalisation abilities.

BackgroundAutomatic sleep staging is essential for assessing sleep quality and diagnosing sleep disorders. While previous research has achieved high classification performance, most current sleep staging networks have only been validated in healthy populations, ignoring the impact of Obstructive Sleep Apnea (OSA) on sleep stage classification. In addition, it remains challenging to effectively improve the fine-grained detection of polysomnography (PSG) and capture multi-scale transitions between sleep stages. Therefore, a more widely applicable network is needed for sleep staging.MethodsThis paper introduces MSDC-SSNet, a novel deep learning network for automatic sleep stage classification. MSDC-SSNet transforms two channels of electroencephalogram (EEG) and one channel of electrooculogram (EOG) signals into time-frequency representations to obtain feature sequences at different temporal and frequency scales. An improved Transformer encoder architecture ensures temporal consistency and effectively captures long-term dependencies in EEG and EOG signals. The Multi-Scale Feature Extraction Module (MFEM) employs convolutional layers with varying dilation rates to capture spatial patterns from fine to coarse granularity. It adaptively fuses the weights of features to enhance the robustness of the model. Finally, multiple channel data are integrated to address the heterogeneity between different modalities effectively and alleviate the impact of OSA on sleep stages.ResultsWe evaluated MSDC-SSNet on three public datasets and our collection of PSG records of 17 OSA patients. It achieved an accuracy of 80.4% on the OSA dataset. It also outperformed the state-of-the-art methods in terms of accuracy, F1 score, and Cohen's Kappa coefficient on the remaining three datasets.ConclusionThe MSDC-SSRNet multi-channel sleep staging architecture proposed in this study enhances widespread system applicability by supplementing inter-channel features. It employs multi-scale attention to extract transition rules between sleep stages and effectively integrates multimodal information. Our method address the limitations of single-channel approaches, enhancing interpretability for clinical applications.

Abstract To monitor in situ the temperature, pH, and micro-strain change information of sandstone artifacts in the process of oxalic acid corrosion, the temperature, pH, and micro-strain fiber Bragg grating (FBG) sensors are developed. A theoretical model of the sensors is established. The surface morphology, material composition, pore structure, temperature, pH, and micro-strain of sandstone corroded by oxalic acid solutions of different concentrations are investigated. The experimental results show that the higher the concentration of oxalic acid, the stronger the corrosiveness of the sandstone. Ferrous oxalate and calcium oxalate produced by corrosion continuously precipitate from the inside of the rock, and the dual reaction of crystallization and dissolution occurs, reducing the proportion of fine pores in the rock pore suction and gravity. The oxalic acid solution is transported to the middle of the sandstone (3 cm) and undergoes the strongest chemical reaction with ore particles and debris, resulting in the maximum wavelength drift of the temperature, pH, and micro-strain sensors. The results of this study provide important support for analyzing the acid dissolution mechanism of stone cultural relics and the preventive protection of cultural relics.

Copper ion (Cu2+) pollution poses a serious threat to marine ecology and fisheries. However, the complexity of seawater and its interference factors make the online detection of Cu2+ quite challenging. To address this issue, we introduce the concept of the photo-assisted adjustment barrier effect into electrochemical detection, using it as a driving force to generate electrochemical responses. The Schottky barrier demonstrates a remarkable regulatory influence on the electrochemical response under photoexcitation, facilitating the response through Cu2+ adsorption. We developed a 4-MBA-AuNPs/CdS/FTO composite that serves as a sensitive platform for Cu2+ detection, achieving a detection limit of 70 nM. Notably, the photo-assisted adjustment of the barrier effect effectively counters the interference posed by ions in seawater, ensuring accurate detection. Furthermore, the sensor exhibits a promising recovery rate (99.62–104.9%) in real seawater samples, highlighting its practical applications. This innovative approach utilizing the photo-assisted adjustment barrier effect offers a promising path for developing electrochemical sensors that can withstand interference.

This work proposes a highly sensitive sandwich heterostructure multimode optical fiber microbend sensor for heart rate (HR), respiratory rate (RR), and ballistocardiography (BCG) monitoring, which is fabricated by combining a sandwich heterostructure multimode fiber Mach–Zehnder interferometer (SHMF-MZI) with a microbend deformer. The parameters of the SHMF-MZI sensor and the microbend deformer were analyzed and optimized in detail, and then the new encapsulated method of the wearable device was put forward. The proposed wearable sensor could greatly enhance the response to the HR signal. The performances for HR, RR, and BCG monitoring were as good as those of the medically approved commercial monitors. The sensor has the advantages of high sensitivity, easy fabrication, and good stability, providing the potential for application in the field of daily supervision and health monitoring.

Abstract In order to realize the intelligent online yield estimation of tomato in the plant factory with artificial lighting (PFAL), a recognition method of tomato red fruit and green fruit based on improved yolov3 deep learning model was proposed to count and estimate tomato fruit yield under natural growth state. According to the planting environment and facility conditions of tomato plants, a computer vision system for fruit counting and yield estimation was designed and the new position loss function was based on the generalized intersection over union (GIoU), which improved the traditional YOLO algorithm loss function. Meanwhile, the scale invariant feature could promote the description precision of the different shapes of fruits. Based on the construction and labeling of the sample image data, the K-means clustering algorithm was used to obtain nine prior boxes of different specifications which were assigned according to the hierarchical level of the feature map. The experimental results of model training and evaluation showed that the mean average precision (mAP) of the improved detection model reached 99.3%, which was 2.7% higher than that of the traditional YOLOv3 model, and the processing time for a single image declined to 15 ms. Moreover, the improved YOLOv3 model had better identification effects for dense and shaded fruits. The research results can provide yield estimation methods and technical support for the research and development of intelligent control system for planting fruits and vegetables in plant factories, greenhouses and fields.

IntroductionIn recent years, innovative sustainable agricultural production technologies, including vertical farms and plant factories, have been developing rapidly around the world. The development of plant factories and vertical farms is currently receiving a lot of attention from Chinese academia and industry. However, the recognition and satisfaction of the government, producers, sales companies, and consumers are low, and their attitudes are mixed, mainly due to high pre-construction and post-operation costs, low comprehensive utilization of resources, low product diversity, low market share, high prices, and low core competitiveness, which limit its healthy and sustainable development. This paper designed a questionnaire from the perspectives of industrialization, commercialization, and sustainability in order to understand the respondents' level of awareness, doubts and concerns, purchase intentions and consumption expectations, as well as their trust and recognition of branded products in this new agricultural production system.MethodTo determine the diversity of respondents, this paper examines the group structure of participants from management, research and development, manufacturing, sales, consumers, regional development, and other industrial sectors. The survey was conducted using both face-to-face interviews and electronic questionnaires. As an online survey, the questionnaire was distributed to social groups through social media platforms. A total of 729 valid questionnaires were submitted. For the purposes of categorizing, enumerating, compiling statistics, and analyzing the questionnaires used in this study, descriptive statistics, multi-factor cross-analysis, and other statistical methods were employed. The social roles, functions, and interaction styles of various plant factory practitioners are examined from the perspective of social relationships, and pertinent development concepts and suggestions are proposed based on the survey results.Results and conclusionsThe study found that an increasing number of consumers are understanding and accepting this new form of plant production and are willing to purchase plant products from plant factories and vertical farming. Plant factories and vertical farms are widely regarded as one of the most important methods of future urban agricultural production. Awareness, purchase intent, price expectations, brand awareness, and price expectations of plant factories and vertical farms varied significantly by gender, age, education level, occupation, and income. In addition, there are numerous findings that provide governments, producers, marketers, managers, and consumers with great value and assistance.Development recommendationsWe should take the opportunity of developing plant factories to adjust the structure of the plant industry, enrich the “vegetable basket” of urban residents, increase the supply capacity of the market, enhance agricultural modernization and technological innovation, improve the quality of agricultural products from plant factories, strengthen the brand sales of plant products, and develop more functional plant products with high added value. Through the development of plant factories and vertical farms, we can improve the nutrition and healthy diet structure of citizens' diets, increase the modern plant industry's contribution to the national economy, and promote the comprehensive and sustainable development of the urban productive plant industry.

This study introduces a new wearable fiber-optic sensor glove. The glove utilizes a flexible material, polydimethylsiloxane (PDMS), and a silicone tube to encapsulate fiber Bragg gratings (FBGs). It is employed to enable the self-perception of hand posture, gesture recognition, and the prediction of grasping objects. The investigation employs the Support Vector Machine (SVM) approach for predicting grasping objects. The proposed fiber-optic sensor glove can concurrently monitor the motion of 14 hand joints comprising 5 metacarpophalangeal joints (MCP), 5 proximal interphalangeal joints (PIP), and 4 distal interphalangeal joints (DIP). To expand the measurement range of the sensors, a sinusoidal layout incorporates the FBG array into the glove. The experimental results indicate that the wearable sensing glove can track finger flexion within a range of 0° to 100°, with a modest minimum measurement error (Error) of 0.176° and a minimum standard deviation (SD) of 0.685°. Notably, the glove accurately detects hand gestures in real-time and even forecasts grasping actions. The fiber-optic smart glove technology proposed herein holds promising potential for industrial applications, including object grasping, 3D displays via virtual reality, and human–computer interaction.

Toxoplasma gondii (T. gondii) is a zoonotic parasite that is widely distributed and seriously endangers public health and human health. Therefore, accurate and effective detection of T. gondii is crucial. This study proposes a microfluidic biosensor using a thin-core microfiber (TCMF) coated with molybdenum disulfide (MoS2) for immune detection of T. gondii. The single-mode fiber was fused with the thin-core fiber, and the TCMF was obtained by arc discharging and flame heating. In order to avoid interference and protect the sensing structure, the TCMF was encapsulated in the microfluidic chip. MoS2 and T. gondii antigen were modified on the surface of TCMF for the immune detection of T. gondii. Experimental results showed that the detection range of the proposed biosensor for T. gondii monoclonal antibody solutions was 1 pg/mL to 10 ng/mL with sensitivity of 3.358 nm/log(mg/mL); the detection of limit was calculated to be 87 fg/mL through the Langmuir model; the dissociation constant and the affinity constant were calculated to be about 5.79 × 10−13 M and 1.727 × 1014 M−1, respectively. The specificity and clinical characteristics of the biosensor was explored. The rabies virus, pseudorabies virus, and T. gondii serum were used to confirm the excellent specificity and clinical characteristics of the biosensor, indicating that the proposed biosensor has great application potential in the biomedical field.

Background Dendrobium catenatum is a valuable traditional medicinal herb with high commercial value. D. catenatum stems contain abundant polysaccharides which are one of the main bioactive components. However, although some genes related to the synthesis of the polysaccharides have been reported, more key genes need to be further elucidated. Results In this study, the contents of polysaccharides and mannose in D. catenatum stems at four developmental stages were compared, and the stems’ transcriptomes were analyzed to explore the synthesis mechanism of the polysaccharides. Many genes involved in starch and sucrose metabolisms were identified by KEGG pathway analysis. Further analysis found that sucrose synthase (SUS; EC 2.4.1.13) gene maybe participated in the polysaccharide synthesis. Hence, we further investigated the genomic characteristics and evolution relationships of the SUS family in plants. The result suggested that the SUS gene of D. catenatum (DcSUS) had undergone the expansion characterized by tandem duplication which might be related to the enrichment of the polysaccharides in D. catenatum stems. Moreover, expression analyses of the DcSUS displayed significant divergent patterns in different tissues and could be divided into two main groups in the stems with four developmental stages. Conclusion In general, our results revealed that DcSUS is likely involved in the metabolic process of the stem polysaccharides, providing crucial clues for exploiting the key genes associated with the polysaccharide synthesis.

Due to their rapid development and wide application in modern agriculture, robots, mobile terminals, and intelligent devices have become vital technologies and fundamental research topics for the development of intelligent and precision agriculture. Accurate and efficient target detection technology is required for mobile inspection terminals, picking robots, and intelligent sorting equipment in tomato production and management in plant factories. However, due to the limitations of computer power, storage capacity, and the complexity of the plant factory (PF) environment, the precision of small-target detection for tomatoes in real-world applications is inadequate. Therefore, we propose an improved Small MobileNet YOLOv5 (SM-YOLOv5) detection algorithm and model based on YOLOv5 for target detection by tomato-picking robots in plant factories. Firstly, MobileNetV3-Large was used as the backbone network to make the model structure lightweight and improve its running performance. Secondly, a small-target detection layer was added to improve the accuracy of small-target detection for tomatoes. The constructed PF tomato dataset was used for training. Compared with the YOLOv5 baseline model, the mAP of the improved SM-YOLOv5 model was increased by 1.4%, reaching 98.8%. The model size was only 6.33 MB, which was 42.48% that of YOLOv5, and it required only 7.6 GFLOPs, which was half that required by YOLOv5. The experiment showed that the improved SM-YOLOv5 model had a precision of 97.8% and a recall rate of 96.7%. The model is lightweight and has excellent detection performance, and so it can meet the real-time detection requirements of tomato-picking robots in plant factories.

X-ray fluorescence computed tomography (XFCT) has attracted wide attention due to its ability to simultaneously and nondestructively obtain structural and elemental distribution information within samples. In this paper, we presented an image system based on the pinhole collimator for the polychromatic L-shell XFCT to reduce time consumption and improve the detection limit. First, the imaging system model was expressed by formulas and discretized. Then, two phantoms (A and B) were scanned by numerical simulation and Monte Carlo simulation. Both phantoms with the same diameter (10 mm) and height (10 mm) were cylinders filled with PMMA, and embedded with GNP-loaded cylinders. The phantom A was inserted by six 1.5 mm-diameter cylinders with different Au concentrations ranging from 0.2% to 1.2%. The phantom B was inserted by eight cylinders with the same Au concentration (1%), but a radius ranging from 0.1 mm to 0.8 mm. Finally, the reconstruction of the XFCT images was performed using the method with and without absorption correction, respectively. The feasibility of XFCT system presented in this paper was demonstrated by the numerical simulation and the Monte Carlo simulation. The results show that absorption attenuation can be corrected by the presented method, and the contrast to noise ratio (CNR) is proportional to Au concentration but almost remains unchanged with the radius of GNP-loaded cylinders, which may provide the necessary justification for further optimization of the imaging system.

Semantic segmentation networks focus on the scene parsing of an unrestricted open scene. The typical segmentation architectures are stacks consisting of convolutional layers, which are used to extract semantic features. The feature map dimension is sharply changed at sampling units for most of networks, which ensure effective propagation of the gradient in deep nets. In this article, we proposed a state-of-the-art network model named Fully Convolutional Pyramidal Networks (FC-PRNet), which employs pyramidal residual structure to change the feature map dimension at all convolutional layers. This design is an effective way of improving generalization ability and optimizing parameters, and FC-PRNet could achieve excellent capability of semantic extraction. We used urban scene benchmark CamVid and KITTI dataset to test our network, the experimental results show that FC-PRNet achieves better results without any pre-training or post-treatment module. Moreover, due to smart construction of pyramidal residual structures, FC-PRNet has less parameters than other existing networks trained on these datasets.

Panicle degradation causes severe yield reduction in rice. There are two main types of panicle degradation: apical spikelet abortion and basal degeneration. In this study, we isolated and characterized the apical panicle abortion mutant apical spikelet abortion (asa), which exhibits degeneration and defects in the apical spikelets. This mutant had a pleiotropic phenotype, characterized by reduced plant height, increased tiller number, and decreased pollen fertility. Map-based cloning revealed that OsASA encodes a boric acid channel protein that showed the highest expression in the inflorescence, peduncle, and anther. RNA-seq analysis of the asa mutant vs wild-type (WT) plants revealed that biological processes related to reactive oxygen species (ROS) homeostasis and salicylic acid (SA) metabolism were significantly affected. Furthermore, the asa mutants had an increased SA level and H2O2 accumulation in the young panicles compared to the WT plants. Moreover, the SA level and the expression of OsPAL3, OsPAL4, and OsPAL6 genes (related to SA biosynthesis) were significantly increased under boron-deficient conditions in the asa mutant and in OsASA-knockout plants. Collectively, these results suggest that the boron distribution maintained by OsASA is required for normal panicle development in a process that involves modulating ROS homeostasis and SA biosynthesis.

The turbidity interference caused by suspended particles in water seriously affects the accuracy of ultraviolet-visible spectroscopy in detecting water quality chemical oxygen demand. Based on this, the application of ultraviolet-visible spectroscopy to detect water quality chemical oxygen demand usually requires physical and mathematical methods to correct the spectral baseline interference caused by turbidity. Because of the slow response speed and unstable compensation effect of traditional correction methods, this paper proposes to use a compressed sensing algorithm to perform baseline correction and achieve good results. In the experiment, we selected formazin turbidity solution and sodium oxalate standard solution and carried out the research on the algorithm of turbidity correction for detecting chemical oxygen demand of water quality by ultraviolet-visible spectroscopy. The experiment obtains the absorption spectra of different concentrations of formazine turbidity solutions and the same concentration of sodium oxalate with different turbidity standard solutions at 210∼845 nm and analyzes the nonlinear effect of absorbance on turbidity. This article uses standard solution experiments to explore the compressed sensing theory for turbidity correction, and through the correction of the absorption spectrum of the actual water sample, it verifies the feasibility of the compression theory for turbidity correction. The method effectively corrects the baseline shift or drift of the water quality ultraviolet-visible absorption spectrum caused by suspended particles, while retaining the absorption characteristics of the ultraviolet spectrum, and it can effectively improve the accuracy and accuracy of the ultraviolet-visible spectroscopy water quality chemical oxygen demand detection.

Minimally invasive surgery (MIS) has been the preferred surgery approach owing to its advantages over conventional open surgery. As a major limitation, the lack of tactile perception impairs the ability of surgeons in tissue distinction and maneuvers. Many studies have been reported on industrial robots to perceive various tactile information. However, only force data are widely used to restore part of the surgeon’s sense of touch in MIS. In recent years, inspired by image classification technologies in computer vision, tactile data are represented as images, where a tactile element is treated as an image pixel. Processing raw data or features extracted from tactile images with artificial intelligence (AI) methods, including clustering, support vector machine (SVM), and deep learning, has been proven as effective methods in industrial robotic tactile perception tasks. This holds great promise for utilizing more tactile information in MIS. This review aims to provide potential tactile perception methods for MIS by reviewing literatures on tactile sensing in MIS and literatures on industrial robotic tactile perception technologies, especially AI methods on tactile images.