Original Network Research Articles

Selection of Grid Road Networks Based on Raster Data

In cartography, generalization is a key process used to simplify complex geographic information, making it suitable for display at different scales while maintaining its essential meaning. When representing high-density road networks on a fixed screen area, overcrowding and loss of clarity often occur. To solve these problems, a road selection operation can be applied. However, traditional methods have primarily focused on structured vector road networks, leaving unstructured raster road networks largely unaddressed. This study introduces a novel technique, Adaptive Road Width Selection (ARWS), designed to improve the multiscale visualization of compact road systems using unstructured raster datasets. The ARWS method begins by segmenting the original raster road network into multilevel superpixels of varying sizes, reflecting the road widths, through neighborhood analysis. Next, road superpixel matching and selection are performed based on the minimum angle and maximum distance rules, alongside shortest-path calculations. Finally, redundant intersection pixels are eliminated to generate the selection results. The proposed ARWS method was evaluated using road network data from Shenzhen, China, producing effective multiscale visualization outcomes. Unlike conventional techniques relying on structured vector data, ARWS excels in preserving the semantic attributes, overall structure, local connectivity, and integrity of road networks. It addresses the challenges of multiscale visualization in dense road networks, offering a robust solution for unstructured raster data.

BRA-YOLOv7: improvements on large leaf disease object detection using FasterNet and dual-level routing attention in YOLOv7

Tea leaf diseases are significant causes of reduced quality and yield in tea production. In the Yunnan region, where the climate is suitable for tea cultivation, tea leaf diseases are small, scattered, and vary in scale, making their detection challenging due to complex backgrounds and issues such as occlusion, overlap, and lighting variations. Existing object detection models often struggle to achieve high accuracy in detecting tea leaf diseases. To address these challenges, this paper proposes a tea leaf disease detection model, BRA-YOLOv7, which combines a dual-level routing dynamic sparse attention mechanism for fast identification of tea leaf diseases in complex scenarios. BRA-YOLOv7 incorporates PConv and FasterNet as replacements for the original network structure of YOLOv7, reducing the number of floating-point operations and improving efficiency. In the Neck layer, a dual-level routing dynamic sparse attention mechanism is introduced to enable flexible computation allocation and content awareness, enhancing the model’s ability to capture global information about tea leaf diseases. Finally, the loss function is replaced with MPDIoU to enhance target localization accuracy and reduce false detection cases. Experiments and analysis were conducted on a collected dataset using the Faster R-CNN, YOLOv6, and YOLOv7 models, with Mean Average Precision (mAP), Floating-point Operations (FLOPs), and Frames Per Second (FPS) as evaluation metrics for accuracy and efficiency. The experimental results show that the improved algorithm achieved a 4.8% improvement in recognition accuracy, a 5.3% improvement in recall rate, a 5% improvement in balance score, and a 2.6% improvement in mAP compared to the traditional YOLOv7 algorithm. Furthermore, in external validation, the floating-point operation count decreased by 1.4G, FPS improved by 5.52%, and mAP increased by 2.4%. In conclusion, the improved YOLOv7 model demonstrates remarkable results in terms of parameter quantity, floating-point operation count, model size, and convergence time. It provides efficient lossless identification while balancing recognition accuracy, real-time performance, and model robustness. This has significant implications for adopting targeted preventive measures against tea leaf diseases in the future.

A lightweight MHDI-DETR model for detecting grape leaf diseases

Accurate diagnosis of grape leaf diseases is critical in agricultural production, yet existing detection techniques face challenges in achieving model lightweighting while ensuring high accuracy. In this study, a real-time, end-to-end, lightweight grape leaf disease detection model, MHDI-DETR, based on an improved RT-DETR architecture, is presented to address these challenges. The original residual backbone network was improved using the MobileNetv4 network, significantly reducing the model’s computational requirements and complexity. Additionally, a lightSFPN feature fusion structure is presented, combining the Hierarchical Scale Feature Pyramid Network with the Dilated Reparam Block structure design from the UniRepLKNet network. This structure is designed to overcome the challenges of capturing complex high-level and subtle low-level features, and it uses Efficient Local Attention to focus more efficiently on regions of interest, thereby enhancing the model’s ability to detect complex targets while improving accuracy and inference speed. Finally, the integration of GIou and Focaler-IoU into Focaler-GIoU enhances detection accuracy and convergence speed for small targets by focusing more effectively on both simple and difficult samples. The findings from the experiments suggest that The MHDI-DETR model results in a 56% decrease in parameters and a 49% reduction in floating-point operations, respectively, compared with the RT-DETR model, in terms of accuracy, the model achieved precision rates of 96.9%, 92.6%, and 72.5% for accuracy, mAP50, and mAP50:95, respectively. Compared with the RT-DETR model, these represent improvements of 1.9%, 1.2%, and 1.2%. Overall, the MHDI-DETR model surpasses the RT-DETR and other mainstream detection models in both detection accuracy and degree of lightness, achieving dual optimization in efficiency and accuracy, and providing an efficient technical solution for automated agricultural disease management.

Tomato Leaf Disease Detection based on the Improved YOLOv10n

In view of the difficulty in distinguishing tomato leaf lesions and the poor identification efficiency, an improved disease detection algorithm CGS-YOLOv10n based on YOLOv10n network was proposed. Firstly, the C2f in the global network is replaced by a multi-scale feature fusion C2f_DBB module, which can extract features more effectively. Secondly, SIOU-SoftNMS is partially used in the neck network to accurately deal with the differences between different targets, so as to improve the classification performance. Finally, the GDetect module was used to replace the detection head part of the original head network, so as to improve the accuracy and efficiency of recognition. The results show that compared with the traditional YOLOv10n, the improved CGS-YOLOv10n algorithm has an accuracy of 93.8%, a recall rate of 91.6%, and an average accuracy of 95.8%. The improved model has higher accuracy and stability in the disease detection of tomato leaf images, which can not only locate and identify targets more accurately, but also effectively reduce the rate of missed detection and false detection. The method proposed in this study provides a reliable technical means for the timely prevention and precise management of tomato diseases in agricultural production, which is helpful to promote the healthy development and sustainable development of agricultural industry.

A Research on Two-Stage Facial Occlusion Recognition Algorithm based on CNN

In recent years, pattern recognition has garnered widespread attention, especially in the domain of face recognition. Traditional face recognition methods have certain limitations in unconstrained environments due to factors such as lighting, facial expressions, and poses. Deep learning can be used to address these challenges. This paper proposes a comprehensive approach to face occlusion recognition based on a two-stage Convolutional Neural Network (CNN). Face verification aims at verifying whether two face images belong to the same individual, and it is a more fundamental task compared to face recognition. The process of face recognition essentially involves multiple instances of face verification, sequentially validating different individuals to ultimately determine the corresponding individual for each face. The primary steps in this research include facial detection, image preprocessing, facial landmark localization, facial landmark extraction, feature matching recognition, and 2D image-assisted 3D face reconstruction. A novel two-stage CNN was designed for facial detection and alignment. The first stage of the network is dedicated to the search for facial windows and regressing vector boundaries. The second stage utilizes 2D images to assist in 3D face reconstruction and perform secondary recognition for cases not identified in the first stage. This method demonstrated excellent performance in handling facial occlusions, achieving high accuracy on datasets such as AFW and FDDB. On the test dataset, face recognition accuracy reached 97.3%, surpassing the original network accuracy of 89.1%. This method outperforms traditional algorithms and general CNN approaches. This study achieved efficient face validation and further handling of unrecognized situations, contributing to the enhancement of face recognition system performance.

Revisiting the traffic flow observability problem: A matrix-based model for traffic networks with or without centroid nodes

This study introduces a graph theory-based model that addresses the link flow observability problem in traffic networks by optimizing passive sensor deployment. The model aims to determine the minimal number of sensors and their optimal placement. It constructs a virtual network and uses isomorphic graph theory to map between the original and virtual networks, ensuring consistency in nodes, links, and link directions. Two formulas are proposed to calculate the minimum number of observable links required across different networks, factoring in links, ordinary nodes, centroid nodes, and added links. Key concepts such as chords, cut sets, and loops, along with their matrices, are analyzed. A matrix-based framework is developed to consider flow conservation conditions. Results show that solving the full link flow observability problem using node flow conservation equations yields a fixed number of sensors with non-unique deployment schemes, Additionally, a resource-constrained sensor network optimization (RSNO) model is presented, employing null space projection (NSP) as an objective function to quantify the impact of budget constraints particularly under the condition if all the link flows cannot be observed. Numerical examples demonstrate the RSNO model's applications.

SpaGAN: A spatially-aware generative adversarial network for building generalization in image maps

Building generalization is an essential task in generating multi-scale topographic maps. The progress of deep learning offers a new paradigm to overcome the coordination challenges faced by conventional building generalization algorithms. Some studies have confirmed the feasibility of several original semantic segmentation networks, such as U-Net and its variants and the conditional generative adversarial network (cGAN), for building generalization in image maps. However, they suffer from critical deformation effects, especially for large and geometrically complex buildings. Since learning building generalization essentially means modeling the subtle transformation of building footprints across scales, we argue that the spatial awareness of a neural network, for instance, regarding building size and shape, is crucial to effective learning. Thus, we propose a spatially-aware generative adversarial network, SpaGAN. It takes a representative cGAN, pix2pix, as the backbone, and modifies two modules: In the U-Net-based generator, an atrous spatial pyramid pooling (ASPP) module replaces the conventional convolutional module to extract multi-scale features of buildings of varying sizes and shapes; in the PatchGAN-based discriminator, a signed distance map (SDM) module is used to capture the fine-grained shape difference for discrimination. The proposed network was comprehensively evaluated with a synthetic and a real-world dataset. The results demonstrate that SpaGAN outperforms existing baseline models (U-Net, ResU-Net, pix2pix) for building generalization, particularly in the real-world dataset. The new model can achieve more reasonable aggregation, simplification, and squaring generalization operators.

Synergistic Effect Enables Aqueous Zinc‐Ion Batteries to Operate at High Temperatures

AbstractThe performance of aqueous zinc‐ion batteries (AZIBs) at high temperatures (HT) is severely compromised by active water corrosion, parasitic reactions, and dendrite growth. Herein, zinc trifluoroacetate is introduced at a low concentration (0.2 m), dissolved in triethyl phosphate (TEP)and H2O. The active water is suppressed due to the reconstructed original hydrogen bond network, which helps inhibit parasitic reactions and severe corrosion. Meanwhile, a solid electrolyte interphase (SEI) formed on the zinc anode due to the decomposition of the introduced zinc salt. The high‐tolerance SEI physically separates the electrolyte and anode, reducing the corrosion caused by active water. Moreover, TEP, as a prevalent fire‐retardant cosolvent, can preferentially anchor on the zinc sheet, serving as a shielding buffer layer. TEP is not only reconstructing the structure of the electric double layer (EDL), decreasing the content of active water, but also accelerating the prompt formation of SEI further. As proof of this synergistic effect, the assembled symmetric Zn.

Vehicle Target Detection Using the Improved YOLOv5s Algorithm

This paper explores the application of the YOLOv5s algorithm integrated with the DeepSORT tracking detection algorithm in vehicle target detection, leveraging its advantages in data processing, loss function, network structure, and training strategy. Regarding detection frame regression, adopting Focal-EIOU can improve vehicle detection accuracy by precisely measuring overlap and better handle complex scenarios, enhancing the overall performance. The CoordConv convolution layer with more spatial position information is employed to enhance the original network structure’s convolution layer and improve vehicle positioning accuracy. The principle and effectiveness of the Shuffle Attention mechanism are analyzed and added to the YOLOv5s network structure to enhance training, improve detection accuracy and running speed. And the DeepSORT tracking detection algorithm is designed to achieve high-speed operation and high-accuracy matching in target tracking, enabling efficient and reliable tracking of objects. Simultaneously, the network structure is optimized to enhance algorithmic speed and performance. To meet the requirements of vehicle detection in practical transportation systems, real-world vehicle images are collected as a dataset for model training to achieve accurate vehicle detection. The results show that the accuracy rate P of the improved YOLOv5s algorithm is increased by 0.484%, and mAP_0.5:0.95 reaches 92.221%, with an increase of 1.747%.

Text Classification based on Spiking Neural Network with LSTM Conversion

In recent years, spiking neural network (SNN) have attracted much attention due to their low energy consumption and have achieved remarkable results in the fields of vision and information processing. However, the application of SNNs in the field of natural language processing (NLP) is still relatively small. Given that current popular large-scale language models rely on huge arithmetic and energy consumption, it is of great practical importance to explore SNN-based approaches to implement NLP tasks in a more energy-efficient way. This paper investigates the conversion method from LSTM network to SNN and compares the performance of the original LSTM network with the converted SNN in a text classification task. The paper experimentally compares the accuracy of different LSTM-based models using different methods on the same dataset. The experimental results show that the converted SNN is able to achieve similar performance to the original LSTM network with significantly lower power consumption in text classification tasks on multiple datasets.

Intraflow temporal correlation-based network traffic prediction

Traditional network traffic prediction methods often rely on complex models that extract features by spatiotemporal dependency analysis in network traffic matrices. However, these methods are limited by specific network topologies and lack generalization ability. Furthermore, it is difficult for traditional time series models to capture multiperiod features. In this study, we propose a network traffic prediction method to address the aforementioned issues, where we combine the flow-by-flow mechanism with an improved TimesNet model. The proposed method transforms the original network traffic matrix into origin-destination sequences using a flow-by-flow mechanism and then uses the improved TimesNet model to transform one-dimensional time series into multiperiod two-dimensional tensors. These transformations not only overcome the limitations of network topologies and the lack of one-dimensional time series representation ability but also focus the analysis on the temporal correlations within the flow. The CoTAttention module is introduced to extract the deep features of periodically changing network traffic, thus improving the accuracy of network traffic prediction. We conduct experiments on the Abilene and GÉANT datasets to verify the effectiveness of the proposed network traffic prediction model based on intraflow temporal correlations. Moreover, we compare its performance with other advanced methods. The experimental results show that the proposed method exhibits superior performance in network traffic prediction tasks, providing powerful guidance for network bandwidth allocation and resource optimization.

RJ-TinyViT: an efficient vision transformer for red jujube defect classification

Compared to the surface defect detection of industrial products produced according to specified processes, the detection of surface defects in naturally grown red jujubes poses unique and significant challenges for researchers. The high diversity of surface defects, subtle distinctions from the background, low contrast, varying scales, and the presence of high levels of noise in images are among the factors that greatly amplify the complexity of defect detection tasks. Existing methods show some deficiencies in addressing these issues, mainly due to insufficient feature extraction capabilities and overly complex network structures, leading to limitations in model efficiency and practical application performance. To tackle the challenges associated with red jujube surface defect detection, this study proposes an optimized Tiny Vision Transformer (TinyViT) network structure, named RJ-TinyViT. This method refines the TinyViT-5 m network structure to reduce network burden and introduces an improved Multi-Kernel Block (MK Block) and an improved Mobile Inverted Bottleneck Convolution Block (MBConv Block) to enhance feature extraction capabilities. Additionally, we have integrated the Coordinate Attention (CA) module to enhance the model’s capacity for recognizing and focusing on features of surface defects on red jujubes. Experimental results show that RJ-TinyViT achieved a classification accuracy of 93.38%, marking an improvement of 1.84% over the original TinyViT network. At the same time, its Floating-point Operations (FLOPs) and Parameters (Params) were reduced to 58.97% and 39.84% of the original TinyViT network, respectively. These results not only demonstrate that RJ-TinyViT achieves model lightweighting while maintaining high accuracy but also highlight its value in practical industrial applications.

Digital Twin-Driven Virtual Network Architecture for Enhanced Extended Reality Capabilities

Extended Reality (XR) technology is proposed as a key enabler for the development of immersive applications, thus transforming the way users interact within digital environments. In communication networks, XR will be able to introduce immersive network functions with the potential to enhance the network experience. This paper proposes a comprehensive architectural framework designed to support the deployment of XR Functions (XRFs) within a Digital Twin Network (DTN) environment in order to provide immersive functions of any type, including control and management functions, that are not available in the original network. The proposal can be deployed in relevant fields where digital twins are promising, such as medical applications, surveillance systems, autonomous driving and space communications. The proposed architecture must consider the need for minimal hardware and software resources, along with a network topological configuration with the objective of providing a robust yet lightweight framework, allowing devices with limited computing power, such as smartphones and laptops, to benefit from the enhanced network functionalities provided in the Digital Twin Network, without affecting the original network (ON).

GFS-YOLO11: A Maturity Detection Model for Multi-Variety Tomato

In order to solve the problems that existing tomato maturity detection methods struggle to take into account both common tomato and cherry tomato varieties in complex field environments (such as light change, occlusion, and fruit overlap) and the model size being too large, this paper proposes a lightweight tomato maturity detection model based on improved YOLO11, named GFS-YOLO11. In order to achieve a lightweight network, we propose the C3k2_Ghost module to replace the C3K2 module in the original network, which can ensure a feature extraction capability and reduce model computation. In order to compensate for the potential feature loss caused by the light weight, this paper proposes a feature-refining module (FRM). After embedding each feature extraction module in the trunk network, it improves the feature expression ability of common tomato and cherry tomato in complex field environments by means of depth-separable convolution, multi-scale pooling, and channel attention and spatial attention mechanisms. In addition, in order to further improve the detection ability of the model for tomatoes of different sizes, the SPPFELAN module is also proposed in this paper. In combining the advantages of SPPF and ELAN, multiple parallel SPPF branches are used to extract features of different levels and perform splicing and fusion. To verify the validity of the method, this study constructed a dataset of 1061 images of common and cherry tomatoes, covering tomatoes in six ripened categories. The experimental results show that the performance of the GFS-YOLO11 model is significantly improved compared with the original model; the P, R, mAP50, and MAP50-95 increased by 5.8%, 4.9%, 6.2%, and 5.5%, respectively, and the number of parameters and calculation amount were reduced by 35.9% and 22.5%, respectively. The GFS-YOLO11 model is lightweight while maintaining high precision, can effectively cope with complex field environments, and more conveniently meet the needs of real-time maturity detection of common tomatoes and cherry tomatoes.

A Variant of the Growing Neural Gas Algorithm for the Design of an Electric Vehicle Charger Network

The Growing Neural Gas (GNG) algorithm constitutes an incremental neural network model based on the idea of a Self-Organizing Map (SOM), that is, unsupervised learning algorithms that reduce the dimensionality of datasets by locating similar samples close to each other. The design of an electric vehicle charging network is an essential aspect in the transition towards more sustainable and environmentally friendly mobility. The need to design and implement an efficient network that meets the needs of all users motivates us to propose the use of a model based on GNG-type neural networks for the design of the network in a specific geographical area. In this paper, a variant of this iterative neural network algorithm is used with the objective that, from an initial dataset of points in the plane, it calculates a new simplified dataset with the main characteristic that the final set of points maintains the geometric shape and topology of the original set. To demonstrate the capabilities of the algorithm, it is exemplified in a real case, in which the design of an electric vehicle charging network is proposed. This network is built by applying the algorithm, taking as the original set of points the ones formed by the nodes of the gas station network in the geographical area studied. Several tests of running the algorithm for different sizes of the final dataset are performed, showing the differences between the original network and the computationally generated one.

Research on improving the ranging accuracy of ships with stereo vision through Kalman filter optimization.

The real-time and accurate detection and ranging of ships play a pivotal role in ensuring navigation safety, this study aims to enhance the navigation safety and environmental perception capabilities of inland waterway vessels. In the ship detection stage, addressing challenges such as large parameters, high computational complexity, and poor real-time performance in existing ship detection models, this paper proposes the MS-YOLOv5s ship target detection algorithm. This algorithm, based on YOLOv5s, utilizes the lightweight MobileNetV3-Small network to replace the original YOLOv5s feature extraction backbone network, thereby improving the detection speed. The results indicate that the parameter size of the MS-YOLOv5s model is 3.55M, only 50.49% of YOLOv5s. Achieving a detection rate of 50.28 FPS, the precision is 96.80%, and the mAP is 98.40%, striking a balance between high accuracy and low computational demand. In the depth estimation stage, influenced by the environment, leading to unstable measurement data, this paper proposes a binocular Kalman filter fusion ranging algorithm. The standard deviation of the ranging results is minimized to 6.032μm, which is one order of magnitude smaller than traditional ranging algorithms, significantly enhancing the robustness of the measurement results. Within a distance of 20m from the ship target, the error can be controlled within 3%, showcasing the applicability of the method proposed in this paper in complex inland waterway environments contributes to the enhancement of ships' environmental perception capabilities and navigation safety, holding positive implications for the development of intelligent vessels.

Cloud Network Collaborative SFC Deployment Under Heterogeneous Resource Of Power White Box Network

The construction of new electric power systems has led to a multiplication in the number of access nodes for electric power IoT information communication, accompanied by a deepening trend in business cloudification. However, due to the insufficient resource linkage and regulation capabilities of the electric power communication network on the cloud side, network side, and edge side, coupled with the lack of a convenient and unified cloud-network resource convergence and control mechanism, the existing electric power communication architecture is gradually finding it difficult to fully meet the demand for deterministic multi-service bearer. Therefore, it is necessary to evolve the original network equipment within the network toward a white box model, opening up interfaces for fine-grained sensing and full-area control, in order to provide more accurate and controllable network transmission services. In this paper, we first propose a cloud-network cooperative resource scheduling architecture for the electric power white box network. This architecture establishes a bearer network to achieve high consistency and cooperativeness of cloud-side end computing power through white box switches, uniformly schedules the global resources of the cloud-network convergence network, and improves the network service quality between cloud-side and side-side. Furthermore, we design a multi-dimensional re-source scheduling method for cloud-network synergy in the electric power white box network. This involves constructing a white box network virtualized heterogeneous resource (VHR) model and a heterogeneous resource control flow (HRCF) model. We transform the problem of merged deployment of service function chains (SFCs) into an SFC merger deployment problem by jointly controlling heterogeneous information, communication, storage, and arithmetic resources, along with the white box switch-specific pipelining resources in the white box network. We then carry out merger optimization of the same kind of function chains. Additionally, we propose a G-GS algorithm based on heuristic methods. Simulation results demonstrate that the method presented in this paper can significantly reduce the cost consumption of white box pipeline resources in the electric power white box network, thereby reducing the processing delay of SFC and improving the quality of electric power service bearing.

Surface defect detection model of laser cutting polycrystalline cubic boron nitride tool based on asymptotic fusion strategy

Due to the polycrystalline cubic boron nitride tool has the characteristics of high hardness, brittleness, etc., it is easy to break the tool or produce defects in the laser cutting process, which affects the cutting performance of the tool. Traditional defect detection methods can no longer meet the needs of modern manufacturing. Aiming at the problems of low accuracy and poor real-time detection of surface defects on laser-cutting polycrystalline cubic boron nitride tools, this study proposes the surface defect detection model of laser cutting polycrystalline cubic boron nitride tool based on asymptotic fusion strategy, which fills the gap in the field. In the backbone network, the C3SE module is constructed by modeling the correlation between feature channels to improve the model’s focus on key features in order to enhance the feature extraction and processing capability of the backbone network; In the neck network, adaptive spatial fusion operation and direct interaction of non-adjacent layers are utilized for multi-scale information fusion, and the asymptotic feature pyramid network for object detection (AFPN) is used instead of the FPN structure to improve the detection performance; In the head network, a soft suppression mechanism is introduced to reduce the overlapping frame score using a decay function, thus improving the detection accuracy. The experimental results based on the self-constructed laser-cutting polycrystalline cubic boron nitride tool surface defects dataset show that the average accuracy of the AFFS-YOLO model is improved by 5.6% compared with that of the YOLOv5 model, reaching 86.1%, and the detection effect is better than that of the original network and other classical target detection networks.

Remote sensing image segmentation of gully erosion in a typical black soil area in Northeast China based on improved DeepLabV3+ model

Gully erosion, a cause of soil degradation and reduced crop yields is widespread in the diffuse river and hillock regions of Northeast China. The surface conditions in gully erosion areas are complex, and manual extraction methods are inefficient, hindering gully erosion monitoring progress. Accurate extraction of gullies is essential for effective environmental monitoring and management. In this study, we propose an improved deep learning network, named DD-DA (the DeepLabV3+ deep network with dual attention mechanism), for automatic and precise gully erosion extraction from 2 m resolution Gaofen-6 remote sensing images from the Heshan Farm area. The DD-DA model builds upon the DeepLabV3+ network architecture with targeted optimizations. Firstly, ResNet50 replaces the original backbone network of DeepLabV3+, enhancing feature extraction capabilities. Second, the Convolutional Block Attention Module (CBAM) integrates spatial and channel attention, improving the model's capacity to capture the shape and texture features of gully erosion. Finally, the Dice loss function is adopted to address the limitations of the traditional cross-entropy loss function, particularly in handling unbalanced datasets. To promote deep learning applications in gully erosion monitoring in Northeast China, a gully erosion dataset was constructed. Experiments on gully erosion datasets show that DD-DA can achieve high Mean Intersection over Union (MIoU) (79.21 %), Pixel Accuracy (96.17 %), Recall (87.66 %), F1-Score (86.48 %), and Precision (86.08 %) simultaneously, which significantly outperform traditional image processing methods and recently deep learning networks. Overall, DD-DA efficiently and accurately segment gully erosion in agricultural fields, significantly improving the extraction of gully erosion features from high-resolution remote sensing images, particularly in the black soil regions of northeastern China. Meanwhile, this study offers a solution for the rapid monitoring of gully erosion and provides substantial technical support for extracting other land degradation features, such as soil salinization and desertification.

Flowering Index Intelligent Detection of Spray Rose Cut Flowers Using an Improved YOLOv5s Model

Addressing the current reliance on manual sorting and grading of spray rose cut flowers, this paper proposed an improved YOLOv5s model for intelligent recognition and grading detection of rose color series and flowering index of spray rose cut flowers. By incorporating small-scale anchor boxes and small object feature output, the model enhanced the annotation accuracy and the detection precision for occluded rose flowers. Additionally, a convolutional block attention module attention mechanism was integrated into the original network structure to improve the model’s feature extraction capability. The WIoU loss function was employed in place of the original CIoU loss function to increase the precision of the model’s post-detection processing. Test results indicated that for two types of spray rose cut flowers, Orange Bubbles and Yellow Bubbles, the improved YOLOv5s model achieved an accuracy and recall improvement of 10.2% and 20.0%, respectively. For randomly collected images of spray rose bouquets, the model maintained a detection accuracy of 95% at a confidence threshold of 0.8.