Upgraded Model Research Articles

Ion motion in strongly magnetized cluster laser plasma

Abstract This paper investigates the interaction of high intensity, circularly polarized, short laser pulses with heavy cluster plasma through analytical modeling and numerical simulations. The optimal parameters were found for the generation of several GigaGs quasi-stationary magnetic field by using the upgraded analytical model and detailed 3D PIC simulations. It is confirmed that a field of such strength slows down the thermal expansion of the cluster core in the direction transverse to the laser beam axis, which can be used in nuclear physics. The generated inward shocks produce ion core compression, which is relevant to nuclear fusion. The electrostatic interaction between the rotating laser field, electrons and ions leads to an ion spiral density distribution in the cluster’s transversal plane, which is absent in a cluster irradiated by linearly polarized pulses with the same parameters.

Study Assesses Effects of Carbon Dioxide on Drilling-Fluid Performance in CCS Wells

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 217711, “CCS Well-Control Impact of CO2 on Drilling-Fluid Performance,” by Jan O. Skogestad, SPE, Blandine Feneuil, SPE, and Elie N’Gouamba, SPE, SINTEF, et al. The paper has not been peer reviewed. _ Drilling carbon-dioxide (CO2) storage wells is a task expected to involve handling an influx of CO2 into the drilling fluid. In the study described in the complete paper, experimental characterization of density, rheology, phase envelope, hydrate formation, and drilling-fluid stability are performed on drilling fluids mixed with CO2 for relevant pressures and temperatures. Based on these experiments and verified thermodynamic model descriptions, local models for properties of drilling-fluid/CO2 mixtures were developed and integrated into an existing drilling software suite. Requirements for Enhanced Well-Control Software Well-control-evaluation software based on sophisticated and verified models of the drilling-fluid/gas interaction in the wellbore plays an important role in planning drilling operations. The software also can be used to evaluate the strategy for circulating out a kick and regaining control of the well. When drilling infill wells into CO2 storage sites, this software must be able to process kicks originating from both CO2 and natural gas influx. To achieve this ability, knowledge gaps related to the interaction of CO2 and drilling fluid must be closed. Hydraulic-flow models need to be enhanced with local models that capture the unique properties of CO2 and how it affects drilling-fluid performance. Starting from carefully devised experimental campaigns, the authors developed local models for drilling-fluid/CO2 interaction that were integrated into an existing hydraulic flow model. A well-control-evaluation product for use during planning and execution of infill drilling operations into CO2 storage reservoirs may then be built using the upgraded models. Experimental methods, equipment, and setup are detailed in the complete paper. Experimental Results Bubblepoints. Bubblepoint experiments in the base oil have been performed with CO2 weight content from 4 to 50% of the total weight of the mixture and at room temperature and 50 and 100°C. All measurements have been repeated at least once. At room temperature, CO2 can be dissolved in oil at mass fractions of up to approximately 50%. The bubblepoint increases when the temperature increases (i.e., the solubility decreases). As with base oil, the solubility of oil-based drilling fluid (OBDF) decreases with temperature. The bubblepoint at room temperature with 25% of CO2 by weight of the mixture is approximately 60 bar, roughly the bubblepoint of pure CO2. Therefore, the maximal solubility of CO2 in OBDF is probably below 25%. To compare the measured solubility in base oil and in OBDF, the CO2 content relative to the base oil has been calculated using the OBDF composition provided by the manufacturer.

A privacy-preserving federated learning approach for airline upgrade optimization

PurposeA key issue of making upgrade decisions is to match the most relevant upgrade offers to the right customers at the right time. To optimize upgrade strategies and profitability, companies seek to break “data silos” between themselves and other business partners for a more holistic view of customers' consumption experiences. However, multi-source data fusion may lead to potential privacy leakage. To overcome these two challenges in data silos and privacy protection, this study introduced a privacy-preserving federated learning (FL) approach and explained the process of using FL in modeling airline passengers’ willingness to pay for upgrade offers. Design/methodology/approachFederated learning is a new confidential computing technique that allows companies to train a model cooperatively by exchanging model parameters instead of the actual raw data, which might include customers' privacy sensitive information. Using a case study of a Chinese airline company, this study demonstrated how FL-based upgrade models using multi-source data can be developed to improve the accuracy of predicting customers' willingness to pay for upgrades while preserving customers’ personal data privacy. FindingsComparing with traditional unilateral model using single-source data, the federated logistic regression and SecureBoost models demonstrate better model performance. This indicates that the proposed FL approach can enhance the accuracy of modeling airline passengers' willingness to pay for upgrade offers while preserving passengers’ data privacy. The findings also show that the FL-based models generally took longer running time than the traditional unilateral model due to the design of FL approach in ensuring data privacy. OriginalityThis study contributes to the literature of upgrade optimization by introducing the new FL approach for developing machining learning models to predict customers’ reaction to upgrade offers. Although we focus on the airline industry in our case study, the proposed FL approach can be applied to other industries with a similar issue of upgrade optimization such as hotels or cruise lines, and car rental.

Creation and Preclinical Testing of a Spirometer with Extended Functionality

RELEVANCE. We present the results of preclinical testing of an upgraded portable AR­1 gas flow analyzer produced by JSC Krasnogvardeets (Saint Petersburg) in order to create import-substituting technologies and develop practical and scientific spheres of healthcare.THE AIM OF THE STUDY. was to substantiate the possibility of clinical use of the upgraded device, originally intended for testing ventilators and inhalation anesthesia devices, as a monitor of the external respiration apparatus function.MATERIAL AND METHODS. The preclinical study of the upgraded gas flow analyzer was conducted at the V.A. Almazov National Medical Research Center in comparison with devices for conventional (Bellavista 950, Berner Ross Medical, Russia) and high-frequency (IPV-2C, Intrapulmonary Percussive Ventilation, Percussionaire Corporation, USA) mechanical ventilation in volume- and pressure-controlled modes using an artificial lung. The following reflected spirometry parameters were compared: tidal and minute volume under inhalation and exhalation, respiratory rate, maximum and minimum flows, peak and positive end expiratory pressure, fraction of inspired oxygen.RESULTS. A high accuracy of parameter matching shown by the gas flow analyzer and ventilators was demonstrated.CONCLUSION. The upgraded model of the gas flow analyzer can be used clinically as a portable spirometer at the prehospital and hospital stages.

Development of a clinical prediction model for pathological upgrading in low-grade squamous intraepithelial lesions following cervical conization

Development of a clinical prediction model for pathological upgrading in low-grade squamous intraepithelial lesions following cervical conization

An Efficient Transfer Learning-Based Framework for Health Care Application

Deep learning has revolutionized healthcare applications, particularly in the diagnosis, treatment, and management of infectious diseases. The main objectives of this investigation are to propose several methods for assessing high-resolution X-ray images with the purpose of identifying the occurrence or not of symptoms associated with pneumonia. The objective of this exam was to identify fixes for these existing problems. Our offering entails a deep learning (DL) technique for detecting chest anomalies using the X-ray modality using the EfficientNet B0 model. In order to make accurate diagnoses of pneumonia, both the EfficientNet B0 and the upgraded CNN model undergo extensive data-driven training. The CNN model that underwent upgrades was determined to be the most effective in this analysis because to its high level of accuracy. The results of our research conclusion are that DL models are capable of monitoring pneumonia's development, increasing diagnostic precision overall and giving patients new optimism for immediate relief.

Energy efficient design of buildings in hot climates through cooling of the envelope

The building envelope can afford many enhancements for lowering the cooling load in hot regions. In this paper, an active upgraded model utilizing an evaporative system to cool the envelope will be tested in the relatively humid climate of the coastal area of the Arabian Gulf. A couple of buildings were erected for this purpose. One was fully enclosed by a secondary structure around it, and the gap between the two was cooled using an evaporative cooler. This was compared to the second, which was insulated according to Dubai Municipality standards. A third comparison was also performed, which concerned the standalone secondary envelope with no evaporative cooling. The interior of all three cases was cooled using DX window-type air conditioners, and the power consumption measured. Thermal simulations were performed on similarly designed models using Ecotect 5.5 to see how each system would perform in full-scale functioning buildings. The measured tests on the building models showed a reduction of electrical consumption in both upgrades compared to the regulation insulated reference. The secondary envelope provided savings of 40 % and 47.3 % in the hot months of June and July, while the evaporative solution saved 52.1 % and 53.7 %, respectively. The simulation of a full-size building over the entire year showed the secondary envelope saving 48.9 % and the evaporative assisted 57.15 %.

Discrete macro - models of nonlinear interlocking mechanisms in the out-of-plane failure of masonry walls

AbstractHistorical unreinforced masonry (URM) constructions are generally vulnerable to out-of-plane (OOP) failures due to the absence of rigid floors and poor connections between orthogonal walls. That leads to the activation of rocking mechanisms of external walls, whose ultimate force and displacement are affected by complex nonlinear interactions with sidewalls. These interactions are often neglected in the engineering practice, potentially leading to significant approximations, as demonstrated by experimental and numerical studies available in the literature. As a novel contribution to the field, this paper presents an upgraded discrete macro-element model (DMEM) to predict the rocking capacity of OOP loaded URM walls interacting with sidewalls. Considering both the onset and the evolution of the rocking mechanism of the front wall, interlocking effects with the sidewalls are first simulated through frictional resistances using the macro-block model (MBM) and the nonlinear kinematic approach of limit analysis. Then, the upgraded DMEM is implemented on the basis of the equivalence between the continuous distribution of these forces, introduced as a further novelty of the paper, and the discrete distribution of lateral elastic-plastic links, accounting for mechanical and geometrical nonlinearities. The results of the two models are discussed in terms of both frictional resistance-displacement and pushover curves, referring to a case study of a front wall belonging to a two-storey URM building. The wall response is also compared with the results derived from the original source of the case study and analysed by changing the number of nonlinear links to define different levels of accuracy.

Corporate farming model for sustainable supply chain crude palm oil of independent smallholder farmers

Independent oil palm smallholders in Indonesia contributed significantly to the sustainable of Crude Palm Oil (CPO) supply chain. On the other hand, sustainability in the process of implementing the supply chain system for independent smallholders is an issue that is extremely important for the success of the supply chain system for the CPO company. This study’s objective is to develop an alternative business model with the goal of strengthening the interaction that already exists between independent farmers and cooperatives in ensuring the existence of a supply chain for raw materials and having access to financial resources, knowledge, and technology. The selection of the research location considered several factors, including the size of the area’s oil palm plantations, the number of oil palm farmers who are self-sufficient, and the possibility of locating CPO plant. Study findings that, the new model by business canvas analysis as the upgraded model of existing model business recommended that it should be built based on the cooperative palm oil family which provides farmers with access to financial, replanting, intensification, and extensification and infrastructure by APKASINDO, BPDPKS, Bank, and other stakeholders to sustain the FFB supply chain in cooperation through a partnership program in a long-term scheme. This whole system was developed by corporate farming system based on the cooperative as the off taker of FFB farmers and develop mini plant CPO to produce cooking oil and biofuels and by product as the waste management system as the value added for farmers.

An Expert Model Using Deep Learning for Image-based Pest Identification with the TSLM Approach for Enhancing Precision Farming

The rise of digital agriculture has sparked considerable interest in its potential to revolutionize farming practices by integrating of advanced technologies. Unmanned aerial vehicles (UAVs), commonly known as drones, have emerged as a crucial tool in precision agriculture, offering diverse applications for real-world scenarios. This research focuses on harnessing the power of drones in the context of digital agriculture support, particularly in addressing the challenges of crop protection and pest management. The objective is to develop an innovative approach for crop-disease diagnosis using an upgraded Transfer-Driven Self-Adaptive Learning Model (TSLM) that leverages multispectral remote sensing data of drone-captured images to improve pest classification and streamline pesticide application. The study explores nine potent deep neural network models' capabilities for identifying plant diseases using various approaches within the digital agriculture framework. Transfer learning and advanced feature extraction techniques are employed to tailor these deep neural networks to the specific crop protection context. Using of pre-trained deep learning models for feature extraction and fine-tuning enhances the effectiveness of the proposed model. Evaluating the model's performance, precision, sensitivity, specificity, and F1-score metrics are assessed, leading to the discovery that deep feature extraction and Self-Adaptive Learning Model (SLM) classification outperform traditional transfer learning methods. The novelty of this work lies in its application of communication protocols to coordinate a fleet of drones for crop protection within the digital agriculture framework. By combining deep learning techniques with transfer-driven self-adaptive learning, the proposed approach significantly enhances the accuracy and efficiency of pest classification in precision agriculture. The study's findings offer valuable insights into optimizing drone-based technologies to combat plant disease epidemics, thereby contributing to the advancement of digital agriculture and its role in supporting sustainable farming practices.

Enhancement of beam quality and transmission using phase bunching in the central region of a cyclotron

Phase bunching in the central region of an azimuthally varying field (AVF) cyclotron contributed significantly to the improvement of the extracted beam quality. The phase bunching effect was demonstrated by measurements of the accelerated and extracted beam properties and by a theoretical analysis using the upgraded geometric trajectory model of the AVF cyclotron at the facility of Takasaki Ion accelerators for Advanced Radiation Application. The phase bunching effect on the beam extraction efficiency and emittance was estimated by a comparison between the 260 MeV 20Ne7+ beam of harmonic number h = 2 in the phase-bunching condition (PBC) and the 107 MeV 4He2+ beam of h = 1 in the non-phase-bunching condition (NBC). The PBC narrowed beam phase width and produced near-single-turn extraction, resulting in high extraction efficiency more than twice as high as that for the NBC. It was proved by beam current measurements. Reduction of the beam phase width was confirmed by the theoretical and experimental analysis of correlation between the radial positions at the phase-defining slits and the phase distribution of the particles passing through the slits. The theoretical analysis also showed that the beam emittance before extraction under the PBC is reduced to less than half of the NBC case. Furthermore, the measured normalized beam emittance in the horizontal direction after the extraction for the PBC was remarkably improved to 0.054π mm mrad nearly one-fourth of the emittance of 0.20π mm mrad for the NBC.

Century‐Long Variations in Surface Incident Solar Radiation Over Japan——Characterized by Observations and Reanalyses

ABSTRACTSurface incident solar radiation (Rs) is of great importance in determining the energy balance in the Earth system. In this study, we use century‐long homogenized observations over Japan to constrain five 20th century reanalyses to explore their performance in reproducing Rs variation on different time scales (high‐frequency components [HFCs], for signals with cycles less than 10 years; low‐frequency components [LFCs], for signals with cycles more than 10 years) by ensemble empirical mode decomposition (EEMD) method and to quantify the impact factors on the Rs estimations by the sum of tree (SOT) model. It is found that ERA20C, ERA20CM and CERA20C overestimated Rs by 0.48–1.49 W/m−2, while 20CRv2c and 20CRv3 underestimated Rs by −0.63 and −1.18 W/m−2, respectively for 1931–2010. Poor correlation coefficient (R) was found to be 0.10 for ERA20CM and 0.22 for 20CRv2c. 20CRv2c failed for 1931–1960 but improved considerably for 1961–2010. 20CRv3 uses an upgraded model and assimilates more observations compared with its predecessor 20CRv2c; however, only the original components and HFCs in Rs were improved, with nearly no improvement in the LFCs. In general, CERA20C Rs, with small biases and higher R of 0.73 for original signals, 0.83 for HFCs and LFCs, is superior to other reanalyses. No obvious trend in clear sky Rs demonstrated that reanalysed Rs are insensitive to the aerosol forcings. Therefore, cloud cover and water vapour maybe the main factors influenced reanalysed Rs. Most of time, Rs is more sensitive to cloud cover than vapour pressure for all reanalyses except original signals (with contribution ratio of 0.29 for cloud cover and 0.71 for vapour pressure) and LFCs (with contribution ratio of 0.41 for cloud cover and 0.59 for vapour pressure) in CERA20C, and original signals (with contribution ratio of 0.37 for cloud cover and 0.63 for vapour pressure) in ERA20C. This work pointed out that aerosol related processes such as aerosol forcings or aerosol radiative effect in reanalyses should be improved in the future, which will ultimately improve the interaction between aerosol and cloud in Rs simulations.

A multi-dimensional evaluation of upgrading fenestration systems in single-family dwellings and implications for rebate structures

This paper presents a novel methodology of four interdependent stages to evaluate the economic, environmental, and health benefits of replacing low-energy performing residential windows with higher-performing ones in single-family dwellings. A modeling-based approach is developed to compute and monetize reduction in heating and cooling energy transmittance, greenhouse gas emissions, and premature mortality associated with the decrease in the outdoor fine particulate matter. Linear regression models of energy prices, carbon pricing, and a value per statistical life approach quantify these benefits monetarily. Factors like upgrade type, dwelling window-to-wall ratio, deterioration in energy performance, geographic location, and window placement by orientation are analyzed for their impact on benefits. A case study examining two locations in the province of Ontario, Canada, is carried out. Results show that energy bill saving is sufficient to justify investing in windows with energy-efficient recognition across all dwelling categories in Southern Ontario, though government rebates are necessary for some dwelling types opting to upgrade to the most energy-efficient models. In contrast, rebates are unnecessary in Northern Ontario for any upgraded models due to substantial energy bill savings. The study proposes a per-window rebate policy based on realized benefits, considering various upgrade models, distinct single-detached dwellings, and geographic location eligibility.

The predictive value of coronary artery calcium score combined with traditional risk factors for obstructive coronary heart disease in young people

ObjectivesThis study attempts to compare the predictive effects of several prediction models on obstructive coronary artery disease (OCAD) in young patients (30–50 years old), with a view to providing a new evaluation tool for the prediction of premature coronary artery disease (PCAD).MethodsA total of 532 hospitalized patients aged 30–50 were included in the study.All of them underwent coronary computed tomography angiography (CCTA) for suspected symptoms of coronary heart disease.Coronary artery calcium score (CACS) combined with traditional risk factors and pre-test probability models are the prediction models to be compared in this study.The PTP model was selected from the upgraded Diamond-Forrester model (UDFM) and the Duke clinical score (DCS).ResultsAll patients included in the study were aged 30–50 years. Among them, women accounted for 24.4%, and 355 patients (66.7%) had a CACS of 0. OCAD was diagnosed in 43 patients (8.1%). The CACS combined with traditional risk factors to predict the OCAD area under the curve of receiver operating characteristic (ROC) (AUC = 0.794,p < 0.001) was greater than the PTP models (AUCUDFM=0.6977,p < 0.001;AUCDCS=0.6214,p < 0.001). By calculating the net reclassification index (NRI) and the integrated discrimination index (IDI), the ability to predict the risk of OCAD using the CACS combined with traditional risk factors was improved compared with the PTP models (NRI&IDI > 0,p < 0.05).ConclusionThe predictive value of CACS combined with traditional risk factors for OCAD in young patients is better than the PTP models.

SAIPy: A Python package for single-station earthquake monitoring using deep learning

Seismology has witnessed significant advancements in recent years with the application of deep learning methods to address a broad range of problems. These techniques have demonstrated their remarkable ability to effectively extract statistical properties from extensive datasets, surpassing the capabilities of traditional approaches to an extent. In this study, we present SAIPy, an open-source Python package specifically developed for fast seismic data processing by implementing deep learning. SAIPy offers solutions for multiple seismological tasks, including earthquake signal detection, seismic phase picking, first motion polarity identification and magnitude estimation. We introduce upgraded versions of previously published models such as CREIME_RT capable of identifying earthquakes with an accuracy above 99.8% and a root mean squared error of 0.38 unit in magnitude estimation. These upgraded models outperform state-of-the-art approaches like the Vision Transformer network. SAIPy provides an API that simplifies the integration of these advanced models, including CREIME_RT, DynaPicker_v2, and PolarCAP, along with benchmark datasets. It also, to the best of our knowledge, introduces the first fully automated deep learning based pipeline to process continuous waveforms. The package has the potential to be used for real-time earthquake monitoring to enable timely actions to mitigate the impact of seismic events. Ongoing development efforts aim to further enhance SAIPy’s performance and incorporate additional features that enhance exploration efforts, and it also would be interesting to approach the retraining of the whole package as a multi-task learning problem. A detailed description of all functions is available in a supplementary document.

Real-time prediction of the chemical oxygen demand component parameters in activated sludge model using backpropagation neural network

Activated sludge models are increasingly being adopted to guide the operation of wastewater treatment plants. Chemical oxygen demand (COD) is an indispensable input for such models. To ensure that the activated sludge mathematical model can adapt to various water quality conditions and minimize prediction errors, it is essential to predict the parameters of the COD components in real-time based on the actual influent COD concentrations. However, conventional methods of determining the components' contributions are too intricate and time-consuming to be really useful. In this study, the chemical oxygen demand in the actual waste water treatment plant was disassembled and analyzed. The research involved determining the proportions of each COD component, assessing the reliability of the measurement parameters, and examining potential factors affecting measurement accuracy, including weather conditions, pipeline conditions, and residents' habits. Then, a backpropagation neural network was developed which can deliver real-time predictions for five important contributors to COD in real time. In addition, using the receiver operating characteristics curve and prediction accuracy to evaluate the performance of the prediction model. For all five components, which SS, XS, SI, XA, and XH, the prediction accuracy of model was more than 80 %. The maximum deviation values of these parameters fall within the range of the actual detected values, suggesting that the model's predictions align well with real-world observations, and demonstrated prediction performance adequate for practical application in wastewater treatment. This article can provide research basis for the engineering application of activated sludge model and help for the intelligent upgrading of waste water treatment plants.

Fragility and vulnerability models for poorly-detailed infilled buildings accounting for building-to-building and masonry infill variability

While in poorly-detailed masonry-infilled reinforced concrete (RC) buildings the significance of masonry infills is acknowledged, less attention has been paid to the variability surrounding their mechanical properties and its potential impact on fragility and, consequently, vulnerability models to be used in seismic risk estimation. This study addresses this gap by proposing updated fragility and vulnerability models of existing poorly-detailed masonry infilled and pilotis RC frames that consider the variability in building layout and infill properties. A large building portfolio, representative of those designed according to Italian codes between 1970 and 1980, is selected as case study, incorporating building-to-building variability through various geometrical configurations. Updated modelling uncertainty parameters, specifically derived for existing infilled RC frames, are used, including the variability surrounding the properties of different infill typologies and the occurrence of RC frame shear failure. The portfolio was analysed through multiple-stripe analyses, followed by the development of upgraded fragility models for buildings with different heights and geometrical configurations. Finally, the corresponding vulnerability models were derived using damage-to-loss models calibrated with post-earthquake observation damage data from the 2009 L'Aquila earthquake. The updated fragility and vulnerability curves can be used for more refined seismic risk estimates of large building portfolios, that account for building-to-building and infill-to-infill variability, revised modelling uncertainty estimates and a more robust intensity measure (AvgSa).

Improved Efficiency and Sensitivity Analysis of 3-D Agent-based Model for Pain-related Neural Activity in the Amygdala.

Neuropathic pain is caused by nerve injury and involves brain areas such as the central nucleus of the amygdala (CeA). We developed the first 3-D agent-based model (ABM) of neuropathic pain-related neurons in the CeA using NetLogo3D. The execution time of a single ABM simulation using realistic parameters (e.g., 13,000 neurons and 22,000+ neural connections) is an important factor in the model's usability. In this paper, we describe our efforts to improve the computational efficiency of our 3-D ABM, which resulted in a 28% reduction in execution time on average for a typical simulation. With this upgraded model, we performed one- and two-parameter sensitivity analyses to study the sensitivity of model output to variability in several key parameters along the anterior to posterior axis of the CeA. These results highlight the importance of computational modeling in exploring spatial and cell-type specific properties of brain regions to inform future wet lab experiments.

Enhancing the Performance and Accuracy in Real-Time Football and Player Detection Using Upgraded YOLOv5 Architecture

The study presents a significantly improved version of the YOLOv5 real-time object detection model for football player recognition. The proposed technique includes feature-tuning and hyper-parameter optimization methods that have been carefully selected to enhance both speed and accuracy, resulting in a superior real-time performance of the YOLOv5 architecture. Furthermore, the YOLOv5 model incorporates a SimSPPF module that enables multi-scale feature extraction with less computational power, making it a highly efficient and effective solution. We selected the GhostNet module to reduce complexity and the Slim scale detection layer for precise bounding box prediction. Our tests, conducted with recordings of multiple football matches, demonstrate that our model accurately detects both the football and players even in complex scenarios with occlusions and dynamic illumination. The suggested method outperforms the original YOLOv5n model in terms of precision, recall, and mean average precision at 0.5 IoU. It is also more computationally efficient. This method has potential applications in live broadcasting, player monitoring, and sports analytics. The upgraded YOLOv5 model demonstrates superior accuracy and efficiency compared to previous methods that rely on traditional image processing techniques or two-stage detectors. This makes it highly suitable for practical, real-world deployments.

An Improved Ningxia Desert Herbaceous Plant Classification Algorithm Based on YOLOv8.

Wild desert grasslands are characterized by diverse habitats, uneven plant distribution, similarities among plant class, and the presence of plant shadows. However, the existing models for detecting plant species in desert grasslands exhibit low precision, require a large number of parameters, and incur high computational cost, rendering them unsuitable for deployment in plant recognition scenarios within these environments. To address these challenges, this paper proposes a lightweight and fast plant species detection system, termed YOLOv8s-KDT, tailored for complex desert grassland environments. Firstly, the model introduces a dynamic convolutional KernelWarehouse method to reduce the dimensionality of convolutional kernels and increase their number, thus achieving a better balance between parameter efficiency and representation ability. Secondly, the model incorporates triplet attention into its feature extraction network, effectively capturing the relationship between channel and spatial position and enhancing the model's feature extraction capabilities. Finally, the introduction of a dynamic detection head tackles the issue related to target detection head and attention non-uniformity, thus improving the representation of the target detection head while reducing computational cost. The experimental results demonstrate that the upgraded YOLOv8s-KDT model can rapidly and effectively identify desert grassland plants. Compared to the original model, FLOPs decreased by 50.8%, accuracy improved by 4.5%, and mAP increased by 5.6%. Currently, the YOLOv8s-KDT model is deployed in the mobile plant identification APP of Ningxia desert grassland and the fixed-point ecological information observation platform. It facilitates the investigation of desert grassland vegetation distribution across the entire Ningxia region as well as long-term observation and tracking of plant ecological information in specific areas, such as Dashuikeng, Huangji Field, and Hongsibu in Ningxia.