To address the challenges of low accuracy and insufficient real-time performance in dynamic object detection for UAV surveillance,this paper proposes a novel tracking framework that integrates a lightweight improved Y...To address the challenges of low accuracy and insufficient real-time performance in dynamic object detection for UAV surveillance,this paper proposes a novel tracking framework that integrates a lightweight improved YOLOv5s model with adaptive motion compensation.A UAV-view dynamic feature enhancement strategy is innovatively introduced,and a lightweight detection network combining attention mechanisms and multi-scale fusion is constructed.The robustness of tracking under motion blur scenarios is also optimized.Experimental results demonstrate that the proposed method achieves a mAP@0.5 of 68.2%on the VisDrone dataset and reaches an inference speed of 32 FPS on the NVIDIA Jetson TX2 platform.This significantly improves the balance between accuracy and efficiency in complex scenes,offering reliable technical support for real-time applications such as emergency response.展开更多
A novel dual-branch decoding fusion convolutional neural network model(DDFNet)specifically designed for real-time salient object detection(SOD)on steel surfaces is proposed.DDFNet is based on a standard encoder–decod...A novel dual-branch decoding fusion convolutional neural network model(DDFNet)specifically designed for real-time salient object detection(SOD)on steel surfaces is proposed.DDFNet is based on a standard encoder–decoder architecture.DDFNet integrates three key innovations:first,we introduce a novel,lightweight multi-scale progressive aggregation residual network that effectively suppresses background interference and refines defect details,enabling efficient salient feature extraction.Then,we propose an innovative dual-branch decoding fusion structure,comprising the refined defect representation branch and the enhanced defect representation branch,which enhance accuracy in defect region identification and feature representation.Additionally,to further improve the detection of small and complex defects,we incorporate a multi-scale attention fusion module.Experimental results on the public ESDIs-SOD dataset show that DDFNet,with only 3.69 million parameters,achieves detection performance comparable to current state-of-the-art models,demonstrating its potential for real-time industrial applications.Furthermore,our DDFNet-L variant consistently outperforms leading methods in detection performance.The code is available at https://github.com/13140W/DDFNet.展开更多
Deep learning-based intelligent recognition algorithms are increasingly recognized for their potential to address the labor-intensive challenge of manual pest detection.However,their deployment on mobile devices has b...Deep learning-based intelligent recognition algorithms are increasingly recognized for their potential to address the labor-intensive challenge of manual pest detection.However,their deployment on mobile devices has been constrained by high computational demands.Here,we developed GBiDC-PEST,a mobile application that incorporates an improved,lightweight detection algorithm based on the You Only Look Once(YOLO)series singlestage architecture,for real-time detection of four tiny pests(wheat mites,sugarcane aphids,wheat aphids,and rice planthoppers).GBiDC-PEST incorporates several innovative modules,including GhostNet for lightweight feature extraction and architecture optimization by reconstructing the backbone,the bi-directional feature pyramid network(BiFPN)for enhanced multiscale feature fusion,depthwise convolution(DWConv)layers to reduce computational load,and the convolutional block attention module(CBAM)to enable precise feature focus.The newly developed GBiDC-PEST was trained and validated using a multitarget agricultural tiny pest dataset(Tpest-3960)that covered various field environments.GBiDC-PEST(2.8 MB)significantly reduced the model size to only 20%of the original model size,offering a smaller size than the YOLO series(v5-v10),higher detection accuracy than YOLOv10n and v10s,and faster detection speed than v8s,v9c,v10m and v10b.In Android deployment experiments,GBiDCPEST demonstrated enhanced performance in detecting pests against complex backgrounds,and the accuracy for wheat mites and rice planthoppers was improved by 4.5-7.5%compared with the original model.The GBiDC-PEST optimization algorithm and its mobile deployment proposed in this study offer a robust technical framework for the rapid,onsite identification and localization of tiny pests.This advancement provides valuable insights for effective pest monitoring,counting,and control in various agricultural settings.展开更多
The increasing prevalence of violent incidents in public spaces has created an urgent need for intelligent surveillance systems capable of detecting dangerous objects in real time.While traditional video surveillance ...The increasing prevalence of violent incidents in public spaces has created an urgent need for intelligent surveillance systems capable of detecting dangerous objects in real time.While traditional video surveillance relies on human monitoring,this approach suffers from limitations such as fatigue and delayed response times.This study addresses these challenges by developing an automated detection system using advanced deep learning techniques to enhance public safety.Our approach leverages state-of-the-art convolutional neural networks(CNNs),specifically You Only Look Once version 4(YOLOv4)and EfficientDet,for real-time object detection.The system was trained on a comprehensive dataset of over 50,000 images,enhanced through data augmentation techniques to improve robustness across varying lighting conditions and viewing angles.Cloud-based deployment on Amazon Web Services(AWS)ensured scalability and efficient processing.Experimental evaluations demonstrated high performance,with YOLOv4 achieving 92%accuracy and processing images in 0.45 s,while EfficientDet reached 93%accuracy with a slightly longer processing time of 0.55 s per image.Field tests in high-traffic environments such as train stations and shopping malls confirmed the system’s reliability,with a false alarm rate of only 4.5%.The integration of automatic alerts enabled rapid security responses to potential threats.The proposed CNN-based system provides an effective solution for real-time detection of dangerous objects in video surveillance,significantly improving response times and public safety.While YOLOv4 proved more suitable for speed-critical applications,EfficientDet offered marginally better accuracy.Future work will focus on optimizing the system for low-light conditions and further reducing false positives.This research contributes to the advancement of AI-driven surveillance technologies,offering a scalable framework adaptable to various security scenarios.展开更多
Food waste presents a major global environmental challenge,contributing to resource depletion,greenhouse gas emissions,and climate change.Black Soldier Fly Larvae(BSFL)offer an eco-friendly solution due to their excep...Food waste presents a major global environmental challenge,contributing to resource depletion,greenhouse gas emissions,and climate change.Black Soldier Fly Larvae(BSFL)offer an eco-friendly solution due to their exceptional ability to decompose organic matter.However,accurately identifying larval instars is critical for optimizing feeding efficiency and downstreamapplications,as different stages exhibit only subtle visual differences.This study proposes a real-timemobile application for automatic classification of BSFL larval stages.The systemdistinguishes between early instars(Stages 1–4),suitable for food waste processing and animal feed,and late instars(Stages 5–6),optimal for pupation and industrial use.A baseline YOLO11 model was employed,achieving a mAP50-95 of 0.811.To further improve performance and efficiency,we introduce YOLO11-DSConv,a novel adaptation incorporating Depthwise Separable Convolutions specifically optimized for the unique challenges of BSFL classification.Unlike existing YOLO+DSConv implementations,our approach is tailored for the subtle visual differences between larval stages and integrated into a complete end-to-end system.The enhanced model achieved a mAP50-95 of 0.813 while reducing computational complexity by 15.5%.The proposed system demonstrates high accuracy and lightweight performance,making it suitable for deployment on resource-constrained agricultural devices,while directly supporting circular economy initiatives through precise larval stage identification.By integrating BSFL classification with realtime AI,this work contributes to sustainable food wastemanagement and advances intelligent applications in precision agriculture and circular economy initiatives.Additional supplementary materials and the implementation code are available at the following link:YOLO11-DSConv,Server Side,Mobile Application.展开更多
The advancement of navigation systems for the visually impaired has significantly enhanced their mobility by mitigating the risk of encountering obstacles and guiding them along safe,navigable routes.Traditional appro...The advancement of navigation systems for the visually impaired has significantly enhanced their mobility by mitigating the risk of encountering obstacles and guiding them along safe,navigable routes.Traditional approaches primarily focus on broad applications such as wayfinding,obstacle detection,and fall prevention.However,there is a notable discrepancy in applying these technologies to more specific scenarios,like identifying distinct food crop types or recognizing faces.This study proposes a real-time application designed for visually impaired individuals,aiming to bridge this research-application gap.It introduces a system capable of detecting 20 different food crop types and recognizing faces with impressive accuracies of 83.27%and 95.64%,respectively.These results represent a significant contribution to the field of assistive technologies,providing visually impaired users with detailed and relevant information about their surroundings,thereby enhancing their mobility and ensuring their safety.Additionally,it addresses the vital aspects of social engagements,acknowledging the challenges faced by visually impaired individuals in recognizing acquaintances without auditory or tactile signals,and highlights recent developments in prototype systems aimed at assisting with face recognition tasks.This comprehensive approach not only promises enhanced navigational aids but also aims to enrich the social well-being and safety of visually impaired communities.展开更多
High-resolution vehicular emissions inventories are important for managing vehicular pollution and improving urban air quality. This study developed a vehicular emission inventory with high spatio-temporal resolution ...High-resolution vehicular emissions inventories are important for managing vehicular pollution and improving urban air quality. This study developed a vehicular emission inventory with high spatio-temporal resolution in the main urban area of Chongqing, based on realtime traffic data from 820 RFID detectors covering 454 roads, and the differences in spatiotemporal emission characteristics between inner and outer districts were analysed. The result showed that the daily vehicular emission intensities of CO, hydrocarbons, PM2.5, PM10,and NO_(x) were 30.24, 3.83, 0.18, 0.20, and 8.65 kg/km per day, respectively, in the study area during 2018. The pollutants emission intensities in inner district were higher than those in outer district. Light passenger cars(LPCs) were the main contributors of all-day CO emissions in the inner and outer districts, from which the contributors of NO_(x) emissions were different. Diesel and natural gas buses were major contributors of daytime NO_(x) emissions in inner districts, accounting for 40.40%, but buses and heavy duty trucks(HDTs) were major contributors in outer districts. At nighttime, due to the lifting of truck restrictions and suspension of buses, HDTs become the main NO_(x) contributor in both inner and outer districts,and its three NO_(x) emission peak hours were found, which are different to the peak hours of total NO_(x) emission by all vehicles. Unlike most other cities, bridges and connecting channels are always emission hotspots due to long-time traffic congestion. This knowledge will help fully understand vehicular emissions characteristics and is useful for policymakers to design precise prevention and control measures.展开更多
The new reality of smart distribution systems with use of generation sources of small and medium sizes brings new challenges for the operation of these systems. The complexity and the large number of nodes requires us...The new reality of smart distribution systems with use of generation sources of small and medium sizes brings new challenges for the operation of these systems. The complexity and the large number of nodes requires use of methods which can reduce the processing time of algorithms such as power flow, allowing its use in real time. This paper presents a known methodology for calculating the power flow in three phases using backward/forward sweep method, and also considering other network elements such as voltage regulators, shunt capacitors and sources of dispersed generation of types PV （active power and voltage） and PQ （active and reactive power）. After that, new elements are introduced that allow the parallelization of this algorithm and an adequate distribution of work between the available processors. The algorithm was implemented using a multi-tiered architecture; the processing times were measured in many network configurations and compared with the same algorithm in the serial version.展开更多
A transonic airfoil designed by means of classical point-optimization may result in its dramatically inferior performance under off-design conditions. To overcome this shortcoming, robust design is proposed to find ou...A transonic airfoil designed by means of classical point-optimization may result in its dramatically inferior performance under off-design conditions. To overcome this shortcoming, robust design is proposed to find out the optimal profile of an airfoil to maintain its performance in an uncertain environment. The robust airfoil optimization is aimed to minimize mean values and variances of drag coefficients while satisfying the lift and thickness constraints over a range of Mach numbers. A multi-objective estimation of distribution algorithm is applied to the robust airfoil optimization on the base of the RAE2822 benchmark airfoil. The shape of the airfoil is obtained through superposing ten Hick-Henne shape functions upon the benchmark airfoil. A set of design points is selected according to a uniform design table for aerodynamic evaluation. A Kriging model of drag coefficient is constructed with those points to reduce computing costs. Over the Mach range from 0.7 to 0.8, the airfoil generated by the robust optimization has a configuration characterized by supercritical airfoil with low drag coefficients. The small fluctuation in its drag coefficients means that the performance of the robust airfoil is insensitive to variation of Mach number.展开更多
The phase behavior of gas condensate in reservoir formations differs from that in pressure-volume-temperature(PVT)cells because it is influenced by porous media in the reservoir formations.Sandstone was used as a samp...The phase behavior of gas condensate in reservoir formations differs from that in pressure-volume-temperature(PVT)cells because it is influenced by porous media in the reservoir formations.Sandstone was used as a sample to investigate the influence of porous media on the phase behavior of the gas condensate.The pore structure was first analyzed using computed tomography(CT)scanning,digital core technology,and a pore network model.The sandstone core sample was then saturated with gas condensate for the pressure depletion experiment.After each pressure-depletion state was stable,realtime CT scanning was performed on the sample.The scanning results of the sample were reconstructed into three-dimensional grayscale images,and the gas condensate and condensate liquid were segmented based on gray value discrepancy to dynamically characterize the phase behavior of the gas condensate in porous media.Pore network models of the condensate liquid ganglia under different pressures were built to calculate the characteristic parameters,including the average radius,coordination number,and tortuosity,and to analyze the changing mechanism caused by the phase behavior change of the gas condensate.Four types of condensate liquid(clustered,branched,membranous,and droplet ganglia)were then classified by shape factor and Euler number to investigate their morphological changes dynamically and elaborately.The results show that the dew point pressure of the gas condensate in porous media is 12.7 MPa,which is 0.7 MPa higher than 12.0 MPa in PVT cells.The average radius,volume,and coordination number of the condensate liquid ganglia increased when the system pressure was between the dew point pressure(12.7 MPa)and the pressure for the maximum liquid dropout,Pmax(10.0 MPa),and decreased when it was below Pmax.The volume proportion of clustered ganglia was the highest,followed by branched,membranous,and droplet ganglia.This study provides crucial experimental evidence for the phase behavior changing process of gas condensate in porous media during the depletion production of gas condensate reservoirs.展开更多
Investigation was made into sediment depth at 723 irregularly scattered measurement points which cover all the regions in Taihu Lake, China. The combination of successive correction scheme and geostatistical method wa...Investigation was made into sediment depth at 723 irregularly scattered measurement points which cover all the regions in Taihu Lake, China. The combination of successive correction scheme and geostatistical method was used to get all the values of recent sediment thickness at the 69×69 grids in the whole lake. The results showed that there is the significant difference in sediment depth between the eastern area and the western region, and most of the sediments are located in the western shore-line and northern regimes but just a little in the center and eastern parts. The notable exception is the patch between the center and Xishan Island where the maximum sediment depth is more than 4.0 m. This sediment distribution pattern is more than likely related to the current circulation pattern induced by the prevailing wind-forcing in Taihu Lake. The numerical simulation of hydrodynamics can strong support the conclusion. Sediment effects on water quality was also studied and the results showed that the concentrations of TP, TN and SS in the western part are obviously larger than those in the eastern regime, which suggested that more nutrients can be released from thicker sediment areas.展开更多
With the development of automation in smart grids,network reconfiguration is becoming a feasible approach for improving the operation of distribution systems.A novel reconfiguration strategy was presented to get the o...With the development of automation in smart grids,network reconfiguration is becoming a feasible approach for improving the operation of distribution systems.A novel reconfiguration strategy was presented to get the optimal configuration of improving economy of the system,and then identifying the important nodes.In this strategy,the objectives increase the node importance degree and decrease the active power loss subjected to operational constraints.A compound objective function with weight coefficients is formulated to balance the conflict of the objectives.Then a novel quantum particle swarm optimization based on loop switches hierarchical encoded was employed to address the compound objective reconfiguration problem.Its main contribution is the presentation of the hierarchical encoded scheme which is used to generate the population swarm particles of representing only radial connected solutions.Because the candidate solutions are feasible,the search efficiency would improve dramatically during the optimization process without tedious topology verification.To validate the proposed strategy,simulations are carried out on the test systems.The results are compared with other techniques in order to evaluate the performance of the proposed method.展开更多
We are investigating the distributed optimization problem,where a network of nodes works together to minimize a global objective that is a finite sum of their stored local functions.Since nodes exchange optimization p...We are investigating the distributed optimization problem,where a network of nodes works together to minimize a global objective that is a finite sum of their stored local functions.Since nodes exchange optimization parameters through the wireless network,large-scale training models can create communication bottlenecks,resulting in slower training times.To address this issue,CHOCO-SGD was proposed,which allows compressing information with arbitrary precision without reducing the convergence rate for strongly convex objective functions.Nevertheless,most convex functions are not strongly convex(such as logistic regression or Lasso),which raises the question of whether this algorithm can be applied to non-strongly convex functions.In this paper,we provide the first theoretical analysis of the convergence rate of CHOCO-SGD on non-strongly convex objectives.We derive a sufficient condition,which limits the fidelity of compression,to guarantee convergence.Moreover,our analysis demonstrates that within the fidelity threshold,this algorithm can significantly reduce transmission burden while maintaining the same convergence rate order as its no-compression equivalent.Numerical experiments further validate the theoretical findings by demonstrating that CHOCO-SGD improves communication efficiency and keeps the same convergence rate order simultaneously.And experiments also show that the algorithm fails to converge with low compression fidelity and in time-varying topologies.Overall,our study offers valuable insights into the potential applicability of CHOCO-SGD for non-strongly convex objectives.Additionally,we provide practical guidelines for researchers seeking to utilize this algorithm in real-world scenarios.展开更多
Foreground moving object detection is an important process in various computer vision applications such as intelligent visual surveillance, HCI, object-based video compression, etc. One of the most successful moving o...Foreground moving object detection is an important process in various computer vision applications such as intelligent visual surveillance, HCI, object-based video compression, etc. One of the most successful moving object detection algorithms is based on Adaptive Gaussian Mixture Model （AGMM）. Although ACMM-hased object detection shows very good performance with respect to object detection accuracy, AGMM is very complex model requiring lots of floatingpoint arithmetic so that it should pay for expensive computational cost. Thus, direct implementation of the AGMM-based object detection for embedded DSPs without floating-point arithmetic HW support cannot satisfy the real-time processing requirement. This paper presents a novel rcal-time implementation of adaptive Gaussian mixture model-based moving object detection algorithm for fixed-point DSPs. In the proposed implementation, in addition to changes of data types into fixed-point ones, magnification of the Gaussian distribution technique is introduced so that the integer and fixed-point arithmetic can be easily and consistently utilized instead of real nmnher and floatingpoint arithmetic in processing of AGMM algorithm. Experimental results shows that the proposed implementation have a high potential in real-time applications.展开更多
In automatic visual inspection, the object image subspace should be segmented and matched quickly so that the affine relationship can be built between the template image and the sample image. When the interference is ...In automatic visual inspection, the object image subspace should be segmented and matched quickly so that the affine relationship can be built between the template image and the sample image. When the interference is strong and the illumination is uneven, for example in an industrial application, this can make it difficult to obtain an objects subspace quickly and accurately in real-time. In this paper, a novel strategy is proposed to adopt discrete radial search paths instead of searching all points in an image. Therefore, the searching time can be substantially reduced. In order to reduce the influence coming from the industrial environment, the paper proposes another method that is local energy level set segmentation, which can locate the object subspace more efficiently and accurately. The detection of ＂crown caps＂ is presented as an example in this paper. Detection effects and computing time are compared between several detection methods, and the mechanisms of inspection have also been analyzed.展开更多
A DMVOCC-MVDA （distributed multiversion optimistic concurrency control with multiversion dynamic adjustment） protocol was presented to process mobile distributed real-time transaction in mobile broadcast environment...A DMVOCC-MVDA （distributed multiversion optimistic concurrency control with multiversion dynamic adjustment） protocol was presented to process mobile distributed real-time transaction in mobile broadcast environments. At the mobile hosts, all transactions perform local pre-validation. The local pre-validation process is carried out against the committed transactions at the server in the last broadcast cycle. Transactions that survive in local pre-validation must be submitted to the server for local final validation. The new protocol eliminates conflicts between mobile read-only and mobile update transactions, and resolves data conflicts flexibly by using multiversion dynamic adjustment of serialization order to avoid unnecessary restarts of transactions. Mobile read-only transactions can be committed with no-blocking, and respond time of mobile read-only transactions is greatly shortened. The tolerance of mobile transactions of disconnections from the broadcast channel is increased. In global validation mobile distributed transactions have to do check to ensure distributed serializability in all participants. The simulation results show that the new concurrency control protocol proposed offers better performance than other protocols in terms of miss rate, restart rate, commit rate. Under high work load （think time is ls） the miss rate of DMVOCC-MVDA is only 14.6%, is significantly lower than that of other protocols. The restart rate of DMVOCC-MVDA is only 32.3%, showing that DMVOCC-MVDA can effectively reduce the restart rate of mobile transactions. And the commit rate of DMVOCC-MVDA is up to 61.2%, which is obviously higher than that of other protocols.展开更多
Load distribution is a key technology in hot strip rolling process, which directly influences strip product quality. A multi-objective load distribution model, which takes into account the rolling force margin balance...Load distribution is a key technology in hot strip rolling process, which directly influences strip product quality. A multi-objective load distribution model, which takes into account the rolling force margin balance, roll wear ratio and strip shape control, is presented. To avoid the selection of weight coefficients encountered in single objective optimization, a multi-objective differential evolutionary algorithm, called MaximinDE, is proposed to solve this model. The experimental results based on practical production data indicate that MaximinDE can obtain a good pareto-optimal solution set, which consists of a series of alternative solutions to load distribution. Decision-makers can select a trade-off solution from the pareto-optimal solution set based on their experience or the importance of ob- iectives. In comparison with the empirical load distribution solution, the trade-off solution can achieve a better per- formance, which demonstrates the effectiveness of the multi-objective load distribution optimization. Moreover, the conflicting relationship among different objectives can be also found, which is another advantage of multi-objective load distribution optimization.展开更多
Visual object-tracking is a fundamental task applied in many applications of computer vision. Particle filter is one of the techniques which has been widely used in object tracking. Due to the virtue of extendability ...Visual object-tracking is a fundamental task applied in many applications of computer vision. Particle filter is one of the techniques which has been widely used in object tracking. Due to the virtue of extendability and flexibility on both linear and non-linear environments, various particle filter-based trackers have been proposed in the literature. However, the conventional approach cannot handle very large videos efficiently in the current data intensive information age. In this work, a parallelized particle filter is provided in a distributed framework provided by the Hadoop/Map-Reduce infrastructure to tackle object-tracking tasks. The experiments indicate that the proposed algorithm has a better convergence and accuracy as compared to the traditional particle filter. The computational power and the scalability of the proposed particle filter in single object tracking have been enhanced as well.展开更多
文摘To address the challenges of low accuracy and insufficient real-time performance in dynamic object detection for UAV surveillance,this paper proposes a novel tracking framework that integrates a lightweight improved YOLOv5s model with adaptive motion compensation.A UAV-view dynamic feature enhancement strategy is innovatively introduced,and a lightweight detection network combining attention mechanisms and multi-scale fusion is constructed.The robustness of tracking under motion blur scenarios is also optimized.Experimental results demonstrate that the proposed method achieves a mAP@0.5 of 68.2%on the VisDrone dataset and reaches an inference speed of 32 FPS on the NVIDIA Jetson TX2 platform.This significantly improves the balance between accuracy and efficiency in complex scenes,offering reliable technical support for real-time applications such as emergency response.
基金supported in part by the National Key R&D Program of China(Grant No.2023YFB3307604)the Shanxi Province Basic Research Program Youth Science Research Project(Grant Nos.202303021212054 and 202303021212046)+3 种基金the Key Projects Supported by Hebei Natural Science Foundation(Grant No.E2024203125)the National Science Foundation of China(Grant No.52105391)the Hebei Provincial Science and Technology Major Project(Grant No.23280101Z)the National Key Laboratory of Metal Forming Technology and Heavy Equipment Open Fund(Grant No.S2308100.W17).
文摘A novel dual-branch decoding fusion convolutional neural network model(DDFNet)specifically designed for real-time salient object detection(SOD)on steel surfaces is proposed.DDFNet is based on a standard encoder–decoder architecture.DDFNet integrates three key innovations:first,we introduce a novel,lightweight multi-scale progressive aggregation residual network that effectively suppresses background interference and refines defect details,enabling efficient salient feature extraction.Then,we propose an innovative dual-branch decoding fusion structure,comprising the refined defect representation branch and the enhanced defect representation branch,which enhance accuracy in defect region identification and feature representation.Additionally,to further improve the detection of small and complex defects,we incorporate a multi-scale attention fusion module.Experimental results on the public ESDIs-SOD dataset show that DDFNet,with only 3.69 million parameters,achieves detection performance comparable to current state-of-the-art models,demonstrating its potential for real-time industrial applications.Furthermore,our DDFNet-L variant consistently outperforms leading methods in detection performance.The code is available at https://github.com/13140W/DDFNet.
基金support of the Natural Science Foundation of Jiangsu Province,China(BK20240977)the China Scholarship Council(201606850024)+1 种基金the National High Technology Research and Development Program of China(2016YFD0701003)the Postgraduate Research&Practice Innovation Program of Jiangsu Province,China(SJCX23_1488)。
文摘Deep learning-based intelligent recognition algorithms are increasingly recognized for their potential to address the labor-intensive challenge of manual pest detection.However,their deployment on mobile devices has been constrained by high computational demands.Here,we developed GBiDC-PEST,a mobile application that incorporates an improved,lightweight detection algorithm based on the You Only Look Once(YOLO)series singlestage architecture,for real-time detection of four tiny pests(wheat mites,sugarcane aphids,wheat aphids,and rice planthoppers).GBiDC-PEST incorporates several innovative modules,including GhostNet for lightweight feature extraction and architecture optimization by reconstructing the backbone,the bi-directional feature pyramid network(BiFPN)for enhanced multiscale feature fusion,depthwise convolution(DWConv)layers to reduce computational load,and the convolutional block attention module(CBAM)to enable precise feature focus.The newly developed GBiDC-PEST was trained and validated using a multitarget agricultural tiny pest dataset(Tpest-3960)that covered various field environments.GBiDC-PEST(2.8 MB)significantly reduced the model size to only 20%of the original model size,offering a smaller size than the YOLO series(v5-v10),higher detection accuracy than YOLOv10n and v10s,and faster detection speed than v8s,v9c,v10m and v10b.In Android deployment experiments,GBiDCPEST demonstrated enhanced performance in detecting pests against complex backgrounds,and the accuracy for wheat mites and rice planthoppers was improved by 4.5-7.5%compared with the original model.The GBiDC-PEST optimization algorithm and its mobile deployment proposed in this study offer a robust technical framework for the rapid,onsite identification and localization of tiny pests.This advancement provides valuable insights for effective pest monitoring,counting,and control in various agricultural settings.
文摘The increasing prevalence of violent incidents in public spaces has created an urgent need for intelligent surveillance systems capable of detecting dangerous objects in real time.While traditional video surveillance relies on human monitoring,this approach suffers from limitations such as fatigue and delayed response times.This study addresses these challenges by developing an automated detection system using advanced deep learning techniques to enhance public safety.Our approach leverages state-of-the-art convolutional neural networks(CNNs),specifically You Only Look Once version 4(YOLOv4)and EfficientDet,for real-time object detection.The system was trained on a comprehensive dataset of over 50,000 images,enhanced through data augmentation techniques to improve robustness across varying lighting conditions and viewing angles.Cloud-based deployment on Amazon Web Services(AWS)ensured scalability and efficient processing.Experimental evaluations demonstrated high performance,with YOLOv4 achieving 92%accuracy and processing images in 0.45 s,while EfficientDet reached 93%accuracy with a slightly longer processing time of 0.55 s per image.Field tests in high-traffic environments such as train stations and shopping malls confirmed the system’s reliability,with a false alarm rate of only 4.5%.The integration of automatic alerts enabled rapid security responses to potential threats.The proposed CNN-based system provides an effective solution for real-time detection of dangerous objects in video surveillance,significantly improving response times and public safety.While YOLOv4 proved more suitable for speed-critical applications,EfficientDet offered marginally better accuracy.Future work will focus on optimizing the system for low-light conditions and further reducing false positives.This research contributes to the advancement of AI-driven surveillance technologies,offering a scalable framework adaptable to various security scenarios.
文摘Food waste presents a major global environmental challenge,contributing to resource depletion,greenhouse gas emissions,and climate change.Black Soldier Fly Larvae(BSFL)offer an eco-friendly solution due to their exceptional ability to decompose organic matter.However,accurately identifying larval instars is critical for optimizing feeding efficiency and downstreamapplications,as different stages exhibit only subtle visual differences.This study proposes a real-timemobile application for automatic classification of BSFL larval stages.The systemdistinguishes between early instars(Stages 1–4),suitable for food waste processing and animal feed,and late instars(Stages 5–6),optimal for pupation and industrial use.A baseline YOLO11 model was employed,achieving a mAP50-95 of 0.811.To further improve performance and efficiency,we introduce YOLO11-DSConv,a novel adaptation incorporating Depthwise Separable Convolutions specifically optimized for the unique challenges of BSFL classification.Unlike existing YOLO+DSConv implementations,our approach is tailored for the subtle visual differences between larval stages and integrated into a complete end-to-end system.The enhanced model achieved a mAP50-95 of 0.813 while reducing computational complexity by 15.5%.The proposed system demonstrates high accuracy and lightweight performance,making it suitable for deployment on resource-constrained agricultural devices,while directly supporting circular economy initiatives through precise larval stage identification.By integrating BSFL classification with realtime AI,this work contributes to sustainable food wastemanagement and advances intelligent applications in precision agriculture and circular economy initiatives.Additional supplementary materials and the implementation code are available at the following link:YOLO11-DSConv,Server Side,Mobile Application.
基金supported by theKorea Industrial Technology Association(KOITA)Grant Funded by the Korean government(MSIT)(No.KOITA-2023-3-003)supported by the MSIT(Ministry of Science and ICT),Korea,under the ITRC(Information Technology Research Center)Support Program(IITP-2024-2020-0-01808)Supervised by the IITP(Institute of Information&Communications Technology Planning&Evaluation)。
文摘The advancement of navigation systems for the visually impaired has significantly enhanced their mobility by mitigating the risk of encountering obstacles and guiding them along safe,navigable routes.Traditional approaches primarily focus on broad applications such as wayfinding,obstacle detection,and fall prevention.However,there is a notable discrepancy in applying these technologies to more specific scenarios,like identifying distinct food crop types or recognizing faces.This study proposes a real-time application designed for visually impaired individuals,aiming to bridge this research-application gap.It introduces a system capable of detecting 20 different food crop types and recognizing faces with impressive accuracies of 83.27%and 95.64%,respectively.These results represent a significant contribution to the field of assistive technologies,providing visually impaired users with detailed and relevant information about their surroundings,thereby enhancing their mobility and ensuring their safety.Additionally,it addresses the vital aspects of social engagements,acknowledging the challenges faced by visually impaired individuals in recognizing acquaintances without auditory or tactile signals,and highlights recent developments in prototype systems aimed at assisting with face recognition tasks.This comprehensive approach not only promises enhanced navigational aids but also aims to enrich the social well-being and safety of visually impaired communities.
基金supported by the National Key Research Program(No.2018YFB1601105,No.2018YFB1601102)the Natural Science Foundation of China(No.41975165,No.U1811463)Chongqing Science and Technology Project(No.cstc2019jscxfxydX0035)。
文摘High-resolution vehicular emissions inventories are important for managing vehicular pollution and improving urban air quality. This study developed a vehicular emission inventory with high spatio-temporal resolution in the main urban area of Chongqing, based on realtime traffic data from 820 RFID detectors covering 454 roads, and the differences in spatiotemporal emission characteristics between inner and outer districts were analysed. The result showed that the daily vehicular emission intensities of CO, hydrocarbons, PM2.5, PM10,and NO_(x) were 30.24, 3.83, 0.18, 0.20, and 8.65 kg/km per day, respectively, in the study area during 2018. The pollutants emission intensities in inner district were higher than those in outer district. Light passenger cars(LPCs) were the main contributors of all-day CO emissions in the inner and outer districts, from which the contributors of NO_(x) emissions were different. Diesel and natural gas buses were major contributors of daytime NO_(x) emissions in inner districts, accounting for 40.40%, but buses and heavy duty trucks(HDTs) were major contributors in outer districts. At nighttime, due to the lifting of truck restrictions and suspension of buses, HDTs become the main NO_(x) contributor in both inner and outer districts,and its three NO_(x) emission peak hours were found, which are different to the peak hours of total NO_(x) emission by all vehicles. Unlike most other cities, bridges and connecting channels are always emission hotspots due to long-time traffic congestion. This knowledge will help fully understand vehicular emissions characteristics and is useful for policymakers to design precise prevention and control measures.
文摘The new reality of smart distribution systems with use of generation sources of small and medium sizes brings new challenges for the operation of these systems. The complexity and the large number of nodes requires use of methods which can reduce the processing time of algorithms such as power flow, allowing its use in real time. This paper presents a known methodology for calculating the power flow in three phases using backward/forward sweep method, and also considering other network elements such as voltage regulators, shunt capacitors and sources of dispersed generation of types PV （active power and voltage） and PQ （active and reactive power）. After that, new elements are introduced that allow the parallelization of this algorithm and an adequate distribution of work between the available processors. The algorithm was implemented using a multi-tiered architecture; the processing times were measured in many network configurations and compared with the same algorithm in the serial version.
基金National Natural Science Foundation of China (10377015)
文摘A transonic airfoil designed by means of classical point-optimization may result in its dramatically inferior performance under off-design conditions. To overcome this shortcoming, robust design is proposed to find out the optimal profile of an airfoil to maintain its performance in an uncertain environment. The robust airfoil optimization is aimed to minimize mean values and variances of drag coefficients while satisfying the lift and thickness constraints over a range of Mach numbers. A multi-objective estimation of distribution algorithm is applied to the robust airfoil optimization on the base of the RAE2822 benchmark airfoil. The shape of the airfoil is obtained through superposing ten Hick-Henne shape functions upon the benchmark airfoil. A set of design points is selected according to a uniform design table for aerodynamic evaluation. A Kriging model of drag coefficient is constructed with those points to reduce computing costs. Over the Mach range from 0.7 to 0.8, the airfoil generated by the robust optimization has a configuration characterized by supercritical airfoil with low drag coefficients. The small fluctuation in its drag coefficients means that the performance of the robust airfoil is insensitive to variation of Mach number.
基金the National Natural Science Foundation of China(Nos.52122402,12172334,52034010,52174051)Shandong Provincial Natural Science Foundation(Nos.ZR2021ME029,ZR2022JQ23)Fundamental Research Funds for the Central Universities(No.22CX01001A-4)。
文摘The phase behavior of gas condensate in reservoir formations differs from that in pressure-volume-temperature(PVT)cells because it is influenced by porous media in the reservoir formations.Sandstone was used as a sample to investigate the influence of porous media on the phase behavior of the gas condensate.The pore structure was first analyzed using computed tomography(CT)scanning,digital core technology,and a pore network model.The sandstone core sample was then saturated with gas condensate for the pressure depletion experiment.After each pressure-depletion state was stable,realtime CT scanning was performed on the sample.The scanning results of the sample were reconstructed into three-dimensional grayscale images,and the gas condensate and condensate liquid were segmented based on gray value discrepancy to dynamically characterize the phase behavior of the gas condensate in porous media.Pore network models of the condensate liquid ganglia under different pressures were built to calculate the characteristic parameters,including the average radius,coordination number,and tortuosity,and to analyze the changing mechanism caused by the phase behavior change of the gas condensate.Four types of condensate liquid(clustered,branched,membranous,and droplet ganglia)were then classified by shape factor and Euler number to investigate their morphological changes dynamically and elaborately.The results show that the dew point pressure of the gas condensate in porous media is 12.7 MPa,which is 0.7 MPa higher than 12.0 MPa in PVT cells.The average radius,volume,and coordination number of the condensate liquid ganglia increased when the system pressure was between the dew point pressure(12.7 MPa)and the pressure for the maximum liquid dropout,Pmax(10.0 MPa),and decreased when it was below Pmax.The volume proportion of clustered ganglia was the highest,followed by branched,membranous,and droplet ganglia.This study provides crucial experimental evidence for the phase behavior changing process of gas condensate in porous media during the depletion production of gas condensate reservoirs.
文摘Investigation was made into sediment depth at 723 irregularly scattered measurement points which cover all the regions in Taihu Lake, China. The combination of successive correction scheme and geostatistical method was used to get all the values of recent sediment thickness at the 69×69 grids in the whole lake. The results showed that there is the significant difference in sediment depth between the eastern area and the western region, and most of the sediments are located in the western shore-line and northern regimes but just a little in the center and eastern parts. The notable exception is the patch between the center and Xishan Island where the maximum sediment depth is more than 4.0 m. This sediment distribution pattern is more than likely related to the current circulation pattern induced by the prevailing wind-forcing in Taihu Lake. The numerical simulation of hydrodynamics can strong support the conclusion. Sediment effects on water quality was also studied and the results showed that the concentrations of TP, TN and SS in the western part are obviously larger than those in the eastern regime, which suggested that more nutrients can be released from thicker sediment areas.
基金Project(61102039)supported by the National Natural Science Foundation of ChinaProject(2014AA052600)supported by National Hi-tech Research and Development Plan,China
文摘With the development of automation in smart grids,network reconfiguration is becoming a feasible approach for improving the operation of distribution systems.A novel reconfiguration strategy was presented to get the optimal configuration of improving economy of the system,and then identifying the important nodes.In this strategy,the objectives increase the node importance degree and decrease the active power loss subjected to operational constraints.A compound objective function with weight coefficients is formulated to balance the conflict of the objectives.Then a novel quantum particle swarm optimization based on loop switches hierarchical encoded was employed to address the compound objective reconfiguration problem.Its main contribution is the presentation of the hierarchical encoded scheme which is used to generate the population swarm particles of representing only radial connected solutions.Because the candidate solutions are feasible,the search efficiency would improve dramatically during the optimization process without tedious topology verification.To validate the proposed strategy,simulations are carried out on the test systems.The results are compared with other techniques in order to evaluate the performance of the proposed method.
基金supported in part by the Shanghai Natural Science Foundation under the Grant 22ZR1407000.
文摘We are investigating the distributed optimization problem,where a network of nodes works together to minimize a global objective that is a finite sum of their stored local functions.Since nodes exchange optimization parameters through the wireless network,large-scale training models can create communication bottlenecks,resulting in slower training times.To address this issue,CHOCO-SGD was proposed,which allows compressing information with arbitrary precision without reducing the convergence rate for strongly convex objective functions.Nevertheless,most convex functions are not strongly convex(such as logistic regression or Lasso),which raises the question of whether this algorithm can be applied to non-strongly convex functions.In this paper,we provide the first theoretical analysis of the convergence rate of CHOCO-SGD on non-strongly convex objectives.We derive a sufficient condition,which limits the fidelity of compression,to guarantee convergence.Moreover,our analysis demonstrates that within the fidelity threshold,this algorithm can significantly reduce transmission burden while maintaining the same convergence rate order as its no-compression equivalent.Numerical experiments further validate the theoretical findings by demonstrating that CHOCO-SGD improves communication efficiency and keeps the same convergence rate order simultaneously.And experiments also show that the algorithm fails to converge with low compression fidelity and in time-varying topologies.Overall,our study offers valuable insights into the potential applicability of CHOCO-SGD for non-strongly convex objectives.Additionally,we provide practical guidelines for researchers seeking to utilize this algorithm in real-world scenarios.
基金supported by Soongsil University Research Fund and BK 21 of Korea
文摘Foreground moving object detection is an important process in various computer vision applications such as intelligent visual surveillance, HCI, object-based video compression, etc. One of the most successful moving object detection algorithms is based on Adaptive Gaussian Mixture Model （AGMM）. Although ACMM-hased object detection shows very good performance with respect to object detection accuracy, AGMM is very complex model requiring lots of floatingpoint arithmetic so that it should pay for expensive computational cost. Thus, direct implementation of the AGMM-based object detection for embedded DSPs without floating-point arithmetic HW support cannot satisfy the real-time processing requirement. This paper presents a novel rcal-time implementation of adaptive Gaussian mixture model-based moving object detection algorithm for fixed-point DSPs. In the proposed implementation, in addition to changes of data types into fixed-point ones, magnification of the Gaussian distribution technique is introduced so that the integer and fixed-point arithmetic can be easily and consistently utilized instead of real nmnher and floatingpoint arithmetic in processing of AGMM algorithm. Experimental results shows that the proposed implementation have a high potential in real-time applications.
文摘In automatic visual inspection, the object image subspace should be segmented and matched quickly so that the affine relationship can be built between the template image and the sample image. When the interference is strong and the illumination is uneven, for example in an industrial application, this can make it difficult to obtain an objects subspace quickly and accurately in real-time. In this paper, a novel strategy is proposed to adopt discrete radial search paths instead of searching all points in an image. Therefore, the searching time can be substantially reduced. In order to reduce the influence coming from the industrial environment, the paper proposes another method that is local energy level set segmentation, which can locate the object subspace more efficiently and accurately. The detection of ＂crown caps＂ is presented as an example in this paper. Detection effects and computing time are compared between several detection methods, and the mechanisms of inspection have also been analyzed.
基金Project(20030533011)supported by the National Research Foundation for the Doctoral Program of Higher Education of China
文摘A DMVOCC-MVDA （distributed multiversion optimistic concurrency control with multiversion dynamic adjustment） protocol was presented to process mobile distributed real-time transaction in mobile broadcast environments. At the mobile hosts, all transactions perform local pre-validation. The local pre-validation process is carried out against the committed transactions at the server in the last broadcast cycle. Transactions that survive in local pre-validation must be submitted to the server for local final validation. The new protocol eliminates conflicts between mobile read-only and mobile update transactions, and resolves data conflicts flexibly by using multiversion dynamic adjustment of serialization order to avoid unnecessary restarts of transactions. Mobile read-only transactions can be committed with no-blocking, and respond time of mobile read-only transactions is greatly shortened. The tolerance of mobile transactions of disconnections from the broadcast channel is increased. In global validation mobile distributed transactions have to do check to ensure distributed serializability in all participants. The simulation results show that the new concurrency control protocol proposed offers better performance than other protocols in terms of miss rate, restart rate, commit rate. Under high work load （think time is ls） the miss rate of DMVOCC-MVDA is only 14.6%, is significantly lower than that of other protocols. The restart rate of DMVOCC-MVDA is only 32.3%, showing that DMVOCC-MVDA can effectively reduce the restart rate of mobile transactions. And the commit rate of DMVOCC-MVDA is up to 61.2%, which is obviously higher than that of other protocols.
基金Item Sponsored by National Natural Science Foundation of China(50974039)
文摘Load distribution is a key technology in hot strip rolling process, which directly influences strip product quality. A multi-objective load distribution model, which takes into account the rolling force margin balance, roll wear ratio and strip shape control, is presented. To avoid the selection of weight coefficients encountered in single objective optimization, a multi-objective differential evolutionary algorithm, called MaximinDE, is proposed to solve this model. The experimental results based on practical production data indicate that MaximinDE can obtain a good pareto-optimal solution set, which consists of a series of alternative solutions to load distribution. Decision-makers can select a trade-off solution from the pareto-optimal solution set based on their experience or the importance of ob- iectives. In comparison with the empirical load distribution solution, the trade-off solution can achieve a better per- formance, which demonstrates the effectiveness of the multi-objective load distribution optimization. Moreover, the conflicting relationship among different objectives can be also found, which is another advantage of multi-objective load distribution optimization.
文摘Visual object-tracking is a fundamental task applied in many applications of computer vision. Particle filter is one of the techniques which has been widely used in object tracking. Due to the virtue of extendability and flexibility on both linear and non-linear environments, various particle filter-based trackers have been proposed in the literature. However, the conventional approach cannot handle very large videos efficiently in the current data intensive information age. In this work, a parallelized particle filter is provided in a distributed framework provided by the Hadoop/Map-Reduce infrastructure to tackle object-tracking tasks. The experiments indicate that the proposed algorithm has a better convergence and accuracy as compared to the traditional particle filter. The computational power and the scalability of the proposed particle filter in single object tracking have been enhanced as well.
