Search

Pests have caused significant losses to agriculture, greatly increasing the detection of pests in the planting process and the cost of pest management in the early stages. At this time, advances in computer vision and deep learning for the detection of pests appearing in the crop open the door to the application of target detection algorithms that can greatly improve the efficiency of tomato pest detection and play an important technical role in the realization of the intelligent planting of tomatoes. However, in the natural environment, tomato leaf pests are small in size, large in similarity, and large in environmental variability, and this type of situation can lead to greater detection difficulty. Aiming at the above problems, a network target detection model based on deep learning, YOLONDD, is proposed in this paper. Designing a new loss function, NMIoU (Normalized Wasserstein Distance with Mean Pairwise Distance Intersection over Union), which improves the ability of anomaly processing, improves the model's ability to detect and identify objects of different scales, and improves the robustness to scale changes; Adding a Dynamic head (DyHead) with an attention mechanism will improve the detection ability of targets at different scales, reduce the number of computations and parameters, improve the accuracy of target detection, enhance the overall performance of the model, and accelerate the training process. Adding decoupled head to Head can effectively reduce the number of parameters and computational complexity and enhance the model's generalization ability and robustness. The experimental results show that the average accuracy of YOLONDD can reach 90.1%, which is 3.33% higher than the original YOLOv5 algorithm and is better than SSD, Faster R-CNN, YOLOv7, YOLOv8, RetinaNet, and other target detection networks, and it can be more efficiently and accurately utilized in tomato leaf pest detection. [ABSTRACT FROM AUTHOR]

Target detection algorithms can greatly improve the efficiency of tomato leaf disease detection and play an important technical role in intelligent tomato cultivation. However, there are some challenges in the detection process, such as the diversity of complex backgrounds and the loss of leaf symmetry due to leaf shadowing, and existing disease detection methods have some disadvantages in terms of deteriorating generalization ability and insufficient accuracy. Aiming at the above issues, a target detection model for tomato leaf disease based on deep learning with a global attention mechanism, TDGA, is proposed in this paper. The main idea of TDGA includes three aspects. Firstly, TDGA adds a global attention mechanism (GAM) after up-sampling and down-sampling, as well as in the SPPF module, to improve the feature extraction ability of the target object, effectively reducing the interference of invalid targets. Secondly, TDGA uses a switchable atrous convolution (SAConv) in the C3 module to improve the model's ability to detect. Thirdly, TDGA adopts the efficient IoU loss (EIoU) instead of complete IoU loss (CIoU) to solve the ambiguous definition of aspect ratio and sample imbalance. In addition, the influences of different environmental factors such as single leaf, multiple leaves, and shadows on the performance of tomato disease detection are extensively experimented with and analyzed in this paper, which also verified the robustness of TDGA. The experimental results show that the average accuracy of TDGA reaches 91.40%, which is 2.93% higher than that of the original YOLOv5 network, which is higher than YOLOv5, YOLOv7, YOLOHC, YOLOv8, SSD, Faster R-CNN, RetinaNet and other target detection networks, so that TDGA can be utilized for the detection of tomato leaf disease more efficiently and accurately, even in complex environments. [ABSTRACT FROM AUTHOR]

Due to the dynamic network topology and limit of resources, fault diagnosis for wireless sensor networks is difficult. The existing diagnostic methods consume a lot of communication bandwidth and node resources, which lead to heavy burden of the resources limited network. This paper presents a passive diagnosis method used for fault detection and fault classification based on the time domain features of sensing data (TDSD). Firstly, the feature extraction and analysis of the sensing data are carried out using one-dimensional discrete Gabor transform, and then the data are diagnosed and classified with Self-Organizing Maps (SOM) neural network; finally the current network status and identifying the fault cause are determined. The results show that, comparing with other methods, this method has fewer burdens in network communication, better diagnostic accuracy rate and classification results, and so forth, and it has a high diagnostic accuracy especially for both node fault and network fault. [ABSTRACT FROM AUTHOR]

Deployment is a basic and key issue for all kinds of wireless sensor networks applications. Most of existing researches on deployment problem of wireless sensor network are on generic network level and not for specific application scenarios and also do not utilize any domain knowledge of practical applications. Focusing on the problem of limited number of sensing nodes in soil respiration sensor networks, using domain knowledge such as some lightweight parameters (temperature, humidity, etc.) influencing soil respiration and soil respiration having a day-periodic trend, we proposed a deployment method, TimSim, for soil respiration sensor network based on time domain similarity of lightweight parameters. Lightweight parameters data from positions in the region to be monitored are collected before the deployment of soil respiration sensing nodes, and then time domain similarities of lightweight data among different positions are analyzed, according to which these positions are divided into some groups. A representative position in each group is chosen to deploy a soil respiration sensing node. The experimental results show that TimSim method can place nodes to proper positions so as to monitor regional soil respiration carbon flux effectively with a smaller estimation error than uniform and random deployment methods. [ABSTRACT FROM AUTHOR]

Semantic segmentation of standing trees is important to obtain factors of standing trees from images automatically and effectively. Aiming at the accurate segmentation of multiple standing trees in complex backgrounds, some traditional methods have shortcomings such as low segmentation accuracy and manual intervention. To achieve accurate segmentation of standing tree images effectively, SEMD, a lightweight network segmentation model based on deep learning, is proposed in this article. DeepLabV3+ is chosen as the base framework to perform multi-scale fusion of the convolutional features of the standing trees in images, so as to reduce the loss of image edge details during the standing tree segmentation and reduce the loss of feature information. MobileNet, a lightweight network, is integrated into the backbone network to reduce the computational complexity. Furthermore, SENet, an attention mechanism, is added to obtain the feature information efficiently and suppress the generation of useless feature information. The extensive experimental results show that using the SEMD model the MIoU of the semantic segmentation of standing tree images of different varieties and categories under simple and complex backgrounds reaches 91.78% and 86.90%, respectively. The lightweight network segmentation model SEMD based on deep learning proposed in this paper can solve the problem of multiple standing trees segmentation with high accuracy. [ABSTRACT FROM AUTHOR]

Fast and accurate counting and positioning of flowers is the foundation of automated flower cultivation production. However, it remains a challenge to complete the counting and positioning of high-density flowers against a complex background. Therefore, this paper proposes a lightweight flower counting and positioning model, Light-FC-YOLO, based on YOLOv8s. By integrating lightweight convolution, the model is more portable and deployable. At the same time, a new efficient detection head, Efficient head, and the integration of the LSKA large kernel attention mechanism are proposed to enhance the model's feature detail extraction capability and change the weight ratio of the shallow edge and key point information in the network. Finally, the SIoU loss function with target angle deviation calculation is introduced to improve the model's detection accuracy and target positioning ability. Experimental results show that Light-FC-YOLO, with a model size reduction of 27.2% and a parameter reduction of 39.0%, has a Mean Average Precision (mAP) and recall that are 0.8% and 1.4% higher than YOLOv8s, respectively. In the counting comparison experiment, the coefficient of determination (R2) and Root Mean Squared Error (RMSE) of Light-FC-YOLO reached 0.9577 and 8.69, respectively, both superior to lightweight models such as YOLOv8s. The lightweight flower detection method proposed in this paper can efficiently complete flower positioning and counting tasks, providing technical support and reference solutions for automated flower production management. [ABSTRACT FROM AUTHOR]

Currently, the classification of grapevine black rot disease relies on assessing the percentage of affected spots in the total area, with a primary focus on accurately segmenting these spots in images. Particularly challenging are cases in which lesion areas are small and boundaries are ill-defined, hampering precise segmentation. In our study, we introduce an enhanced U-Net network tailored for segmenting black rot spots on grape leaves. Leveraging VGG as the U-Net's backbone, we strategically position the atrous spatial pyramid pooling (ASPP) module at the base of the U-Net to serve as a link between the encoder and decoder. Additionally, channel and spatial dual-attention modules are integrated into the decoder, alongside a feature pyramid network aimed at fusing diverse levels of feature maps to enhance the segmentation of diseased regions. Our model outperforms traditional plant disease semantic segmentation approaches like DeeplabV3+, U-Net, and PSPNet, achieving impressive pixel accuracy (PA) and mean intersection over union (MIoU) scores of 94.33% and 91.09%, respectively. Demonstrating strong performance across various levels of spot segmentation, our method showcases its efficacy in enhancing the segmentation accuracy of black rot spots on grapevines. [ABSTRACT FROM AUTHOR]

Simple Summary: Blurry scenarios often affect the clarity of fish images, posing significant challenges to deep learning models in terms of the accurate recognition of fish species. A method based on deep learning with a diffusion model and an attention mechanism, DiffusionFR, is proposed herein to improve the accuracy of fish species recognition in blurry scenarios caused by light reflections and water ripple noise. Using a self-constructed dataset, BlurryFish, extensive experiments were conducted and the results showed that the proposed two-stage diffusion network model can restore the clarity of blurry fish images to some extent and the proposed learnable attention module is effective in improving the accuracy of fish species recognition. Blurry scenarios, such as light reflections and water ripples, often affect the clarity and signal-to-noise ratio of fish images, posing significant challenges for traditional deep learning models in accurately recognizing fish species. Firstly, deep learning models rely on a large amount of labeled data. However, it is often difficult to label data in blurry scenarios. Secondly, existing deep learning models need to be more effective for the processing of bad, blurry, and otherwise inadequate images, which is an essential reason for their low recognition rate. A method based on the diffusion model and attention mechanism for fish image recognition in blurry scenarios, DiffusionFR, is proposed to solve these problems and improve the performance of species recognition of fish images in blurry scenarios. This paper presents the selection and application of this correcting technique. In the method, DiffusionFR, a two-stage diffusion network model, TSD, is designed to deblur bad, blurry, and otherwise inadequate fish scene pictures to restore clarity, and a learnable attention module, LAM, is intended to improve the accuracy of fish recognition. In addition, a new dataset of fish images in blurry scenarios, BlurryFish, was constructed and used to validate the effectiveness of DiffusionFR, combining bad, blurry, and otherwise inadequate images from the publicly available dataset Fish4Knowledge. The experimental results demonstrate that DiffusionFR achieves outstanding performance on various datasets. On the original dataset, DiffusionFR achieved the highest training accuracy of 97.55%, as well as a Top-1 accuracy test score of 92.02% and a Top-5 accuracy test score of 95.17%. Furthermore, on nine datasets with light reflection noise, the mean values of training accuracy reached a peak at 96.50%, while the mean values of the Top-1 accuracy test and Top-5 accuracy test were at their highest at 90.96% and 94.12%, respectively. Similarly, on three datasets with water ripple noise, the mean values of training accuracy reached a peak at 95.00%, while the mean values of the Top-1 accuracy test and Top-5 accuracy test were at their highest at 89.54% and 92.73%, respectively. These results demonstrate that the method showcases superior accuracy and enhanced robustness in handling original datasets and datasets with light reflection and water ripple noise. [ABSTRACT FROM AUTHOR]

Ground-object classification using remote-sensing images of high resolution is widely used in land planning, ecological monitoring, and resource protection. Traditional image segmentation technology has poor effect on complex scenes in high-resolution remote-sensing images. In the field of deep learning, some deep neural networks are being applied to high-resolution remote-sensing image segmentation. The DeeplabV3+ network is a deep neural network based on encoder-decoder architecture, which is commonly used to segment images with high precision. However, the segmentation accuracy of high-resolution remote-sensing images is poor, the number of network parameters is large, and the cost of training network is high. Therefore, this paper improves the DeeplabV3+ network. Firstly, MobileNetV2 network was used as the backbone feature-extraction network, and an attention-mechanism module was added after the feature-extraction module and the ASPP module to introduce focal loss balance. Our design has the following advantages: it enhances the ability of network to extract image features; it reduces network training costs; and it achieves better semantic segmentation accuracy. Experiments on high-resolution remote-sensing image datasets show that the mIou of the proposed method on WHDLD datasets is 64.76%, 4.24% higher than traditional DeeplabV3+ network mIou, and the mIou on CCF BDCI datasets is 64.58%. This is 5.35% higher than traditional DeeplabV3+ network mIou and outperforms traditional DeeplabV3+, U-NET, PSP-NET and MACU-net networks. [ABSTRACT FROM AUTHOR]

Estimation of regional soil carbon flux is very important for the study of the global carbon cycle. The spatial heterogeneity of soil respiration prevents the actual status of regional soil carbon flux from being revealed by measurements of only one or a few spatial sampling positions, which are usually used by traditional studies for the limitation of measurement instruments, so measuring in many spatial positions is very necessary. However, the existing instruments are expensive and cannot communicate with each other, which prevents them from meeting the requirement of synchronous measurements in multiple positions. Therefore, we designed and implemented an instrument for soil carbon flux measuring based on dynamic chamber method, SCFSen, which can measure soil carbon flux and communicate with each other to construct a sensor network. In its working stage, a SCFSen node measures the concentration of carbon in the chamber with an infrared carbon dioxide sensor for certain times periodically, and then the changing rate of the measurements is calculated, which can be converted to the corresponding value of soil carbon flux in the position during the short period. A wireless sensor network system using SCFSens as soil carbon flux sensing nodes can carry out multi-position measurements synchronously, so as to obtain the spatial heterogeneity of soil respiration. Furthermore, the sustainability of such a wireless sensor network system makes the temporal variability of regional soil carbon flux can also be obtained. So SCFSen makes thorough monitoring and accurate estimation of regional soil carbon flux become more feasible. [ABSTRACT FROM AUTHOR]

The selective cutting method currently used in Moso bamboo forests has resulted in a reduction of stand productivity and carbon sequestration capacity. Given the time and labor expense involved in addressing this problem manually, simulation using an ecosystem model is the most suitable approach. The BIOME-BGC model was improved to suit managed Moso bamboo forests, which was adapted to include age structure, specific ecological processes and management measures of Moso bamboo forest. A field selective cutting experiment was done in nine plots with three cutting intensities (high-intensity, moderate-intensity and low-intensity) during 2010–2013, and biomass of these plots was measured for model validation. Then four selective cutting scenarios were simulated by the improved BIOME-BGC model to optimize the selective cutting timings, intervals, retained ages and intensities. The improved model matched the observed aboveground carbon density and yield of different plots, with a range of relative error from 9.83% to 15.74%. The results of different selective cutting scenarios suggested that the optimal selective cutting measure should be cutting 30% culms of age 6, 80% culms of age 7, and all culms thereafter (above age 8) in winter every other year. The vegetation carbon density and harvested carbon density of this selective cutting method can increase by 74.63% and 21.5%, respectively, compared with the current selective cutting measure. The optimized selective cutting measure developed in this study can significantly promote carbon density, yield, and carbon sink capacity in Moso bamboo forests. [ABSTRACT FROM AUTHOR]

Numerical models are the most appropriate instrument for the analysis of the carbon balance of terrestrial ecosystems and their interactions with changing environmental conditions. The process-based model BIOME-BGC is widely used in simulation of carbon balance within vegetation, litter and soil of unmanaged ecosystems. For Moso bamboo forests, however, simulations with BIOME-BGC are inaccurate in terms of the growing season and the carbon allocation, due to the oversimplified representation of phenology. Our aim was to improve the applicability of BIOME-BGC for managed Moso bamboo forest ecosystem by implementing several new modules, including phenology, carbon allocation, and management. Instead of the simple phenology and carbon allocation representations in the original version, a periodic Moso bamboo phenology and carbon allocation module was implemented, which can handle the processes of Moso bamboo shooting and high growth during “on-year” and “off-year”. Four management modules (digging bamboo shoots, selective cutting, obtruncation, fertilization) were integrated in order to quantify the functioning of managed ecosystems. The improved model was calibrated and validated using eddy covariance measurement data collected at a managed Moso bamboo forest site (Anji) during 2011–2013 years. As a result of these developments and calibrations, the performance of the model was substantially improved. Regarding the measured and modeled fluxes (gross primary production, total ecosystem respiration, net ecosystem exchange), relative errors were decreased by 42.23%, 103.02% and 18.67%, respectively. [ABSTRACT FROM AUTHOR]

Bamboo forests provide important ecosystem services and play an important role in terrestrial carbon cycling. Of the approximately 500 bamboo species in China, Moso bamboo ( Phyllostachys pubescens ) is the most important one in terms of distribution, timber value, and other economic values. In this study, we estimated current and potential carbon stocks in China's Moso bamboo forests and in their products. The results showed that Moso bamboo forests in China stored about 611.15 ± 142.31 Tg C, 75% of which was in the top 60 cm soil, 22% in the biomass of Moso bamboos, and 3% in the ground layer (i.e., bamboo litter, shrub, and herb layers). Moso bamboo products store 10.19 ± 2.54 Tg C per year. The potential carbon stocks reach 1331.4 ± 325.1 Tg C, while the potential C stored in products is 29.22 ± 7.31 Tg C a −1 . Our results indicate that Moso bamboo forests and products play a critical role in C sequestration. The information gained in this study will facilitate policy decisions concerning carbon sequestration and management of Moso bamboo forests in China. [ABSTRACT FROM AUTHOR]

• Canopy structure parameters had significant correlation with ΔAT in summer. • Different 3D canopy geometry shapes and planting types have different cooling effect in summer. • SR is a key factor affecting PET. • The ΔAT exhibited a logarithmic relationship with the distance from the lake. As an important component of urban forests, roadside trees provide cooling and humidification effects, thereby improving the microclimate and human thermal comfort during hot summer weather by blocking, reflecting and absorbing solar radiation. However, the complex three-dimensional (3D) structure of roadside trees, diverse planting types and complex landscape heterogeneity complicate efforts to quantify the effects of these factors on above-road air temperatures. In this study, a terrestrial laser scanner (TLS) is used to accurately estimate roadside tree morphological parameters, such as the tree height (TH) and the diameter at breast height (DBH), and canopy structural parameters, such as the canopy thickness (CT), the canopy cover above the road (CCAD), canopy volume over the road (CVAD) and the leaf area index (LAI), to evaluate different 3D canopy geometries (pyramidal, spherical, pileate and cylindrical). A moving meteorological station was used to continuously obtain the horizontal air temperature (AT) distribution to investigate the effects of the canopy structural parameters, the planting types and the land cover types on the above-road AT. The value of ΔAT was positively correlated with TH, CVAD, LAI and CCAD. These structural variables were most significantly correlated with ΔAT under the pileate canopy shape. However, there were no significant correlations between these factors under the cylindrical canopy shape. The physiological equivalent temperature (ΔPET) reached 7.4, 15.8, 4.5 and 17 °C under the cylindrical, spherical, pyramidal and pileate shapes, respectively. The ΔAT of a two-sided planting was approximately 1.7 times greater than that of a one-sided planting; the ΔRH showed a similar trend. Within 230 m of a lake, the ΔAT exhibited a logarithmic relationship with the distance from a given point to the lake. Our study quantifies the influence of the roadside tree structure on the thermal environment, which can serve as a meaningful reference for urban tree species selection and arrangement. [ABSTRACT FROM AUTHOR]

Highlights: [•] This study analyzed ice storm impacts on GPP and ET of Moso bamboo forest. [•] GPP and ET were accurately predicted using EC measurements and ecosystem models. [•] Ice storm decreased plant transpiration but increased soil evaporation. [•] This ice storm decreased annual mean GPP by 0.17gCm−2 d−1. [•] Ice storm damage to the off-year was slightly greater than that to the on-year. [Copyright &y& Elsevier]