In modern industrial production,foreign object detection in complex environments is crucial to ensure product quality and production safety.Detection systems based on deep-learning image processing algorithms often fa...In modern industrial production,foreign object detection in complex environments is crucial to ensure product quality and production safety.Detection systems based on deep-learning image processing algorithms often face challenges with handling high-resolution images and achieving accurate detection against complex backgrounds.To address these issues,this study employs the PatchCore unsupervised anomaly detection algorithm combined with data augmentation techniques to enhance the system’s generalization capability across varying lighting conditions,viewing angles,and object scales.The proposed method is evaluated in a complex industrial detection scenario involving the bogie of an electric multiple unit(EMU).A dataset consisting of complex backgrounds,diverse lighting conditions,and multiple viewing angles is constructed to validate the performance of the detection system in real industrial environments.Experimental results show that the proposed model achieves an average area under the receiver operating characteristic curve(AUROC)of 0.92 and an average F1 score of 0.85.Combined with data augmentation,the proposed model exhibits improvements in AUROC by 0.06 and F1 score by 0.03,demonstrating enhanced accuracy and robustness for foreign object detection in complex industrial settings.In addition,the effects of key factors on detection performance are systematically analyzed,providing practical guidance for parameter selection in real industrial applications.展开更多
Transorbital craniocerebral injury is a relatively rare type of penetrating head injury that poses a significant threat to the ocular and cerebral structures.^([1])The clinical prognosis of transorbital craniocerebral...Transorbital craniocerebral injury is a relatively rare type of penetrating head injury that poses a significant threat to the ocular and cerebral structures.^([1])The clinical prognosis of transorbital craniocerebral injury is closely related to the size,shape,speed,nature,and trajectory of the foreign object,as well as the incidence of central nervous system damage and secondary complications.The foreign objects reported to have caused these injuries are categorized into wooden items,metallic items,^([2-8])and other materials,which penetrate the intracranial region via fi ve major pathways,including the orbital roof (OR),superior orbital fissure (SOF),inferior orbital fissure(IOF),optic canal (OC),and sphenoid wing.Herein,we present eight cases of transorbital craniocerebral injury caused by an unusual metallic foreign body.展开更多
Aiming at the limitations of the existing railway foreign object detection methods based on two-dimensional(2D)images,such as short detection distance,strong influence of environment and lack of distance information,w...Aiming at the limitations of the existing railway foreign object detection methods based on two-dimensional(2D)images,such as short detection distance,strong influence of environment and lack of distance information,we propose Rail-PillarNet,a three-dimensional(3D)LIDAR(Light Detection and Ranging)railway foreign object detection method based on the improvement of PointPillars.Firstly,the parallel attention pillar encoder(PAPE)is designed to fully extract the features of the pillars and alleviate the problem of local fine-grained information loss in PointPillars pillars encoder.Secondly,a fine backbone network is designed to improve the feature extraction capability of the network by combining the coding characteristics of LIDAR point cloud feature and residual structure.Finally,the initial weight parameters of the model were optimised by the transfer learning training method to further improve accuracy.The experimental results on the OSDaR23 dataset show that the average accuracy of Rail-PillarNet reaches 58.51%,which is higher than most mainstream models,and the number of parameters is 5.49 M.Compared with PointPillars,the accuracy of each target is improved by 10.94%,3.53%,16.96%and 19.90%,respectively,and the number of parameters only increases by 0.64M,which achieves a balance between the number of parameters and accuracy.展开更多
The location of an ingested foreign object is often difficult to determine by X-ray if gastric air bubbles are not clear in the image.Methods that provide negative contrast can facilitate precise object localization,w...The location of an ingested foreign object is often difficult to determine by X-ray if gastric air bubbles are not clear in the image.Methods that provide negative contrast can facilitate precise object localization,which is important for object retrieval and treatment of the patient.This case report describes a male child,2 years and 2 mo of age,who accidentally swallowed a lithium battery while playing at home.A plain X-ray showed that the battery was in the abdomen,but it was unclear whether the object was still inside the stomach.A second X-ray examination performed after oral administration of a bloating agent to produce expansion of the stomach and provide negative contrast confirmed that the ingested battery was still in the stomach.The battery was then carefully removed using magnetic and balloon catheters under fluoroscopic guidance.This case report describes the successful use of an orally administered bloating agent without pain to the child in orderto determine the precise location of a foreign object in the abdomen.展开更多
To address the issues of small target miss detection,false positives in complex scenarios,and insufficient real-time performance in maglev train foreign object intrusion detection,this paper proposes a multi-module fu...To address the issues of small target miss detection,false positives in complex scenarios,and insufficient real-time performance in maglev train foreign object intrusion detection,this paper proposes a multi-module fusion improvement algorithm,YOLO11-FADA(Fusion of Augmented Features and Dynamic Attention),based on YOLO11.The model achieves collaborative optimization through three key modules:The Local Feature Augmentation Module(LFAM)enhances small target features and mitigates feature loss during down-sampling through multi-scale feature parallel extraction and attention fusion.The Dynamically Tuned Self-Attention(DTSA)module introduces learnable parameters to adjust attentionweights dynamically,and,in combinationwith convolution,expands the receptive field to suppress complex background interference.TheWeighted Convolution 2D(wConv2D)module optimizes convolution kernel weights using symmetric density functions and sparsification,reducing the parameter count by 30% while retaining core feature extraction capabilities.YOLO11-FADA achieves a mAP@0.5 of 0.907 on a custom maglev train foreign object dataset,improving by 3.0% over the baseline YOLO11 model.The model’s computational complexity is 7.3 GFLOPs,with a detection speed of 118.6 FPS,striking a balance between detection accuracy and real-time performance,thereby offering an efficient solution for rail transit safety monitoring.展开更多
With the rapid development and widespread application of electric vehicles(EVs)around the world,the wireless power transfer(WPT)technology is also accelerating for commercial applications in EV wireless charging(EV-WP...With the rapid development and widespread application of electric vehicles(EVs)around the world,the wireless power transfer(WPT)technology is also accelerating for commercial applications in EV wireless charging(EV-WPT)because of its high reliability,safety,and convenience,especially high suitability for the future self-driving scenario.Foreign object detection(FOD),mainly including metal object detection and living object detection,is required urgently and timely for the practical application of EV-WPT technology to ensure electromagnetic safety.In the last decade,especially in the past three years,many pieces of research on FOD have been reported.This article reviews FOD state-of-the-art technology for EV-WPT and compares the pros and cons of different approaches in terms of sensitivity,reliability,adaptability,complexity,and cost.Future challenges for research and development are also discussed to encourage commercialisation of EV-WPT technique.展开更多
Objects in agricultural soils will seriously affect the farming operations of agricultural machinery.At present,it still relies on human experience to judge abnormal Gounrd-penetrting Radar(GPR)signals.It is difficult...Objects in agricultural soils will seriously affect the farming operations of agricultural machinery.At present,it still relies on human experience to judge abnormal Gounrd-penetrting Radar(GPR)signals.It is difficult for traditional image processing technology to form a general positioning method for the randomness and diversity characteristics of GPR signals in soil.Although many scholars had researched a variety of image-processing techniques,most methods lack robustness.In this study,the deep learning algorithm Mask Region-based Convolutional Neural Network(Mask-RCNN)and a geometric model were combined to improve the GPR positioning accuracy.First,a soil stratification experiment was set to classify the physical parameters of the soil and study the attenuation law of electromagnetic waves.Secondly,a SOIL-GPR geometric model was proposed,which can be combined with Mask-RCNN's MASK geometric size to predict object sizes.The results proved the effectiveness and accuracy of the model for position detection and evaluation of objects in soils;then,the improved Mask RCNN method was used to compare the feature extraction accuracy of U-Net and Fully Convolutional Networks(FCN);Finally,the operating speed of agricultural machinery was simulated and designed the A-B survey line experiment.The detection accuracy was evaluated by several indicators,such as the survey line direction,soil depth false alarm rate,Mean Average Precision(mAP),and Intersection over Union(IoU).The results showed that pixel-level segmentation and positioning based on Mask RCNN can improve the accuracy of the position detection of objects in agricultural soil effectively,and the average error of depth prediction is 2.87 cm.The results showed that the detection technology proposed in this study integrates the advantage of soil environmental parameters,geometric models,and artificial intelligence algorithms to provide a high-precision and technical solution for the GPR non-destructive detection of soils.展开更多
[Objective]As wireless power transfer(WPT)technology is increasingly deployed in scenarios such as electric vehicles,metallic foreign objects in the WPT area may cause local overheating and energy loss.Existing method...[Objective]As wireless power transfer(WPT)technology is increasingly deployed in scenarios such as electric vehicles,metallic foreign objects in the WPT area may cause local overheating and energy loss.Existing methods still suffer from poor edge/corner sensitivity,misjudgment due to fixed thresholds,and limited ability to extract position information.This work proposes a wireless power transfer-foreign object detection(WPT-FOD)method based on channel differential response and a dynamic-threshold corner-enhancement strategy,aiming to improve detection sensitivity,localization accuracy,and robustness without altering the overall coil layout.[Method]A multi-channel detection coil array is designed,and the self-inductance disturbance response of each channel coil is modeled.A channel-difference mapping mechanism is introduced to build a 2-D sensitivity matrix to characterize spatial position correlation.A corner-enhancement algorithm is developed to weight and amplify the collaborative response of adjacent channels,and a dynamic threshold adjustment mechanism is integrated to adapt to varying interference levels.Validation is carried out on a self-built 64-channel FOD platform by moving a typical metallic foreign object across central,edge,and corner regions,and by conducting comparative tests under different interference intensities.[Result]When a typical metallic foreign object moves to corner regions,the self-inductance disturbance of the detection coil increases from less than 0.02μH to more than 0.06μH,significantly enhancing the discrimination capability at corners.Under varying interference strengths,the dynamic threshold mechanism reduces the number of false positives from 13 to 2,demonstrating good environmental adaptability and stability.[Conclusion]By combining channel differential response,corner enhancement,and dynamic thresholding,the proposed WPT-FOD effectively mitigates edge/corner blind spots and fixed-threshold misjudgment,while providing localization capability and robustness.It markedly improves the accuracy of metallic foreign object detection in WPT systems and offers a feasible path and method reference for the safe application and engineering deployment of WPT systems.展开更多
文摘In modern industrial production,foreign object detection in complex environments is crucial to ensure product quality and production safety.Detection systems based on deep-learning image processing algorithms often face challenges with handling high-resolution images and achieving accurate detection against complex backgrounds.To address these issues,this study employs the PatchCore unsupervised anomaly detection algorithm combined with data augmentation techniques to enhance the system’s generalization capability across varying lighting conditions,viewing angles,and object scales.The proposed method is evaluated in a complex industrial detection scenario involving the bogie of an electric multiple unit(EMU).A dataset consisting of complex backgrounds,diverse lighting conditions,and multiple viewing angles is constructed to validate the performance of the detection system in real industrial environments.Experimental results show that the proposed model achieves an average area under the receiver operating characteristic curve(AUROC)of 0.92 and an average F1 score of 0.85.Combined with data augmentation,the proposed model exhibits improvements in AUROC by 0.06 and F1 score by 0.03,demonstrating enhanced accuracy and robustness for foreign object detection in complex industrial settings.In addition,the effects of key factors on detection performance are systematically analyzed,providing practical guidance for parameter selection in real industrial applications.
文摘Transorbital craniocerebral injury is a relatively rare type of penetrating head injury that poses a significant threat to the ocular and cerebral structures.^([1])The clinical prognosis of transorbital craniocerebral injury is closely related to the size,shape,speed,nature,and trajectory of the foreign object,as well as the incidence of central nervous system damage and secondary complications.The foreign objects reported to have caused these injuries are categorized into wooden items,metallic items,^([2-8])and other materials,which penetrate the intracranial region via fi ve major pathways,including the orbital roof (OR),superior orbital fissure (SOF),inferior orbital fissure(IOF),optic canal (OC),and sphenoid wing.Herein,we present eight cases of transorbital craniocerebral injury caused by an unusual metallic foreign body.
基金supported by a grant from the National Key Research and Development Project(2023YFB4302100)Key Research and Development Project of Jiangxi Province(No.20232ACE01011)Independent Deployment Project of Ganjiang Innovation Research Institute,Chinese Academy of Sciences(E255J001).
文摘Aiming at the limitations of the existing railway foreign object detection methods based on two-dimensional(2D)images,such as short detection distance,strong influence of environment and lack of distance information,we propose Rail-PillarNet,a three-dimensional(3D)LIDAR(Light Detection and Ranging)railway foreign object detection method based on the improvement of PointPillars.Firstly,the parallel attention pillar encoder(PAPE)is designed to fully extract the features of the pillars and alleviate the problem of local fine-grained information loss in PointPillars pillars encoder.Secondly,a fine backbone network is designed to improve the feature extraction capability of the network by combining the coding characteristics of LIDAR point cloud feature and residual structure.Finally,the initial weight parameters of the model were optimised by the transfer learning training method to further improve accuracy.The experimental results on the OSDaR23 dataset show that the average accuracy of Rail-PillarNet reaches 58.51%,which is higher than most mainstream models,and the number of parameters is 5.49 M.Compared with PointPillars,the accuracy of each target is improved by 10.94%,3.53%,16.96%and 19.90%,respectively,and the number of parameters only increases by 0.64M,which achieves a balance between the number of parameters and accuracy.
文摘The location of an ingested foreign object is often difficult to determine by X-ray if gastric air bubbles are not clear in the image.Methods that provide negative contrast can facilitate precise object localization,which is important for object retrieval and treatment of the patient.This case report describes a male child,2 years and 2 mo of age,who accidentally swallowed a lithium battery while playing at home.A plain X-ray showed that the battery was in the abdomen,but it was unclear whether the object was still inside the stomach.A second X-ray examination performed after oral administration of a bloating agent to produce expansion of the stomach and provide negative contrast confirmed that the ingested battery was still in the stomach.The battery was then carefully removed using magnetic and balloon catheters under fluoroscopic guidance.This case report describes the successful use of an orally administered bloating agent without pain to the child in orderto determine the precise location of a foreign object in the abdomen.
文摘To address the issues of small target miss detection,false positives in complex scenarios,and insufficient real-time performance in maglev train foreign object intrusion detection,this paper proposes a multi-module fusion improvement algorithm,YOLO11-FADA(Fusion of Augmented Features and Dynamic Attention),based on YOLO11.The model achieves collaborative optimization through three key modules:The Local Feature Augmentation Module(LFAM)enhances small target features and mitigates feature loss during down-sampling through multi-scale feature parallel extraction and attention fusion.The Dynamically Tuned Self-Attention(DTSA)module introduces learnable parameters to adjust attentionweights dynamically,and,in combinationwith convolution,expands the receptive field to suppress complex background interference.TheWeighted Convolution 2D(wConv2D)module optimizes convolution kernel weights using symmetric density functions and sparsification,reducing the parameter count by 30% while retaining core feature extraction capabilities.YOLO11-FADA achieves a mAP@0.5 of 0.907 on a custom maglev train foreign object dataset,improving by 3.0% over the baseline YOLO11 model.The model’s computational complexity is 7.3 GFLOPs,with a detection speed of 118.6 FPS,striking a balance between detection accuracy and real-time performance,thereby offering an efficient solution for rail transit safety monitoring.
基金Key R&D Program of Guangdong Province,China(No.2020B0404030004)partly by the open research fund from Guangdong Laboratory of Artificial Intelligence and Digital Economy(SZ)(No.GML-KF-22-19)partly by the National Natural Science Foundation of China(No.62001301).
文摘With the rapid development and widespread application of electric vehicles(EVs)around the world,the wireless power transfer(WPT)technology is also accelerating for commercial applications in EV wireless charging(EV-WPT)because of its high reliability,safety,and convenience,especially high suitability for the future self-driving scenario.Foreign object detection(FOD),mainly including metal object detection and living object detection,is required urgently and timely for the practical application of EV-WPT technology to ensure electromagnetic safety.In the last decade,especially in the past three years,many pieces of research on FOD have been reported.This article reviews FOD state-of-the-art technology for EV-WPT and compares the pros and cons of different approaches in terms of sensitivity,reliability,adaptability,complexity,and cost.Future challenges for research and development are also discussed to encourage commercialisation of EV-WPT technique.
基金supported by the Laboratory of Lingnan Modern Agriculture Project(Grant No.NT2021009)Guangdong University Key Field(Artificial Intelligence)Special Project(No.2019KZDZX1012)and the 111 Project(D18019)+3 种基金Guangdong Basic and Applied Basic Research Foundation(Grant No.2021A1515110554)China Postdoctoral Science Foundation(Grant No.2022M721201)the National Natural Science Foundation of China(Grant No.31901411)The Open Competition Program of the Top Ten Critical Priorities of Agricultural Science and Technology Innovation for the 14th Five-Year Plan of Guangdong Province(No.2022SDZG03).
文摘Objects in agricultural soils will seriously affect the farming operations of agricultural machinery.At present,it still relies on human experience to judge abnormal Gounrd-penetrting Radar(GPR)signals.It is difficult for traditional image processing technology to form a general positioning method for the randomness and diversity characteristics of GPR signals in soil.Although many scholars had researched a variety of image-processing techniques,most methods lack robustness.In this study,the deep learning algorithm Mask Region-based Convolutional Neural Network(Mask-RCNN)and a geometric model were combined to improve the GPR positioning accuracy.First,a soil stratification experiment was set to classify the physical parameters of the soil and study the attenuation law of electromagnetic waves.Secondly,a SOIL-GPR geometric model was proposed,which can be combined with Mask-RCNN's MASK geometric size to predict object sizes.The results proved the effectiveness and accuracy of the model for position detection and evaluation of objects in soils;then,the improved Mask RCNN method was used to compare the feature extraction accuracy of U-Net and Fully Convolutional Networks(FCN);Finally,the operating speed of agricultural machinery was simulated and designed the A-B survey line experiment.The detection accuracy was evaluated by several indicators,such as the survey line direction,soil depth false alarm rate,Mean Average Precision(mAP),and Intersection over Union(IoU).The results showed that pixel-level segmentation and positioning based on Mask RCNN can improve the accuracy of the position detection of objects in agricultural soil effectively,and the average error of depth prediction is 2.87 cm.The results showed that the detection technology proposed in this study integrates the advantage of soil environmental parameters,geometric models,and artificial intelligence algorithms to provide a high-precision and technical solution for the GPR non-destructive detection of soils.
文摘[Objective]As wireless power transfer(WPT)technology is increasingly deployed in scenarios such as electric vehicles,metallic foreign objects in the WPT area may cause local overheating and energy loss.Existing methods still suffer from poor edge/corner sensitivity,misjudgment due to fixed thresholds,and limited ability to extract position information.This work proposes a wireless power transfer-foreign object detection(WPT-FOD)method based on channel differential response and a dynamic-threshold corner-enhancement strategy,aiming to improve detection sensitivity,localization accuracy,and robustness without altering the overall coil layout.[Method]A multi-channel detection coil array is designed,and the self-inductance disturbance response of each channel coil is modeled.A channel-difference mapping mechanism is introduced to build a 2-D sensitivity matrix to characterize spatial position correlation.A corner-enhancement algorithm is developed to weight and amplify the collaborative response of adjacent channels,and a dynamic threshold adjustment mechanism is integrated to adapt to varying interference levels.Validation is carried out on a self-built 64-channel FOD platform by moving a typical metallic foreign object across central,edge,and corner regions,and by conducting comparative tests under different interference intensities.[Result]When a typical metallic foreign object moves to corner regions,the self-inductance disturbance of the detection coil increases from less than 0.02μH to more than 0.06μH,significantly enhancing the discrimination capability at corners.Under varying interference strengths,the dynamic threshold mechanism reduces the number of false positives from 13 to 2,demonstrating good environmental adaptability and stability.[Conclusion]By combining channel differential response,corner enhancement,and dynamic thresholding,the proposed WPT-FOD effectively mitigates edge/corner blind spots and fixed-threshold misjudgment,while providing localization capability and robustness.It markedly improves the accuracy of metallic foreign object detection in WPT systems and offers a feasible path and method reference for the safe application and engineering deployment of WPT systems.
