Efficient data processing on embedded devices may reduce network communication and improve battery usage allowing for longer sensor lifetime. Data processing is challenged by limited CPU and memory hardware. EmbedDB i...
详细信息
With people's constant pursuit of a better life, population problems, food problems and health problems are becoming increasingly prominent, which hastens the development of Plant factory technology. China's P...
详细信息
A comprehensive approach to energy-efficient smart home automation leveraging IoT technologies. The Altair Smart Core IoT platform integrates with Arduino, enabling remote control and management of household devices. ...
详细信息
Taking a single-phase three-level ANPC off-grid inverter as an example, a current estimation method based on state observer is proposed in this paper. The topology structure, mathematical model, proposed current estim...
详细信息
Surface defect detection of workpieces plays an important role in qualitycontrol of workpieces. Traditional manual detection methods are inefficient, costly, and prone to problems such as false positives and false ne...
详细信息
ISBN:
(纸本)9798350352573
Surface defect detection of workpieces plays an important role in qualitycontrol of workpieces. Traditional manual detection methods are inefficient, costly, and prone to problems such as false positives and false negatives. To address this, a modified YOLOv5 object detection model for surface defect detection is proposed, aiming to reduce inefficiencies and issues with false positives and false negatives. Firstly, the CBAM attention mechanism is introduced to enhance the network's perception ability, increase the feature weights of object detection, and suppress the influence of irrelevant features. Secondly, the unique residual module bottleneck in the C3 module is replaced with the Res2Net residual module to enhance the network's feature extraction capability and increase the receptive field size. Finally, the MPDIoU loss function is used to better optimize the accuracy and stability of object detection during training, further improving the quality of detection results. Experimental results show that the improved model achieves an mAP value of 77.7% on the NEU-DET dataset, which is 4.8% higher than the YOLOv5 network, demonstrating a significant improvement in detection performance.
The proceedings contain 202 papers. The special focus in this conference is on Chinese Intelligent systems. The topics include: Modeling of Interval Type-2 Fuzzy Logic System by Semi-tensor Product;Robust Ad...
ISBN:
(纸本)9789819786497
The proceedings contain 202 papers. The special focus in this conference is on Chinese Intelligent systems. The topics include: Modeling of Interval Type-2 Fuzzy Logic System by Semi-tensor Product;Robust Adaptive control of Missiles Based on Fuzzy RBF Neural network;deep Reinforcement Learning of Physically Simulated Character control;the Game of Degree of Target-Attacker-Defender Game With Non-zero Capture Radius;Visual Inertial 3D Reconstruction System Based on KLT Optical Flow and Voxel Hash;analysis of Students’ Class Status Based on Deep Learning;fault-Tolerant Attitude control of Hypersonic Flight Vehicles Based on Extended Adaptive Iterative Learning;predefined-Time control of Robot systems with Prescribed performance Guarantees;dynamic Coverage of Unicycle Robots with Anisotropic Sensing for Time-Priority;method for Evaluating the Multi-factor Decision-Making Efficiency Based on Dynamic Bayesian network;submarine Cable Target Search Based on Improved Biological Neural network;end-to-End mmWave-Based Human Pose Estimation from Raw Signal;urban Sewage Treatment Plant Simulation System Based on Industrial internet;Load Matching Design and Multi-objective Optimization of Helicopter EHA;Testability Design and Verification of Typical Functional Links of CNI System;adaptive Sliding Mode Tracking control for Multiple High-Speed Trains with Actuator Saturation and Parameter Uncertainties;Abnormal Data Detection Based on Dual-Factor Weighted SVDD for Multimode Batch Processes;dual-Branch Attention Dense Residual network Based on Negative Image for Image Denoising;Research on Testability Work Planning Technology for Aviation Equipment Based on DoDAF;aero Engine Instability Prediction and Detection Method Based on Gated Recurrent Neural networks;Seven-Degrees-of-Freedom Robotic Arm Path Planning Based on Improved RRT;dynamic Event-Triggered Based Coherent Formation control for Multi-robot systems;adaptive Prescribed Time Prescribed performancecontrol for High-Order No
The proceedings contain 202 papers. The special focus in this conference is on Chinese Intelligent systems. The topics include: Modeling of Interval Type-2 Fuzzy Logic System by Semi-tensor Product;Robust Ad...
ISBN:
(纸本)9789819786534
The proceedings contain 202 papers. The special focus in this conference is on Chinese Intelligent systems. The topics include: Modeling of Interval Type-2 Fuzzy Logic System by Semi-tensor Product;Robust Adaptive control of Missiles Based on Fuzzy RBF Neural network;deep Reinforcement Learning of Physically Simulated Character control;the Game of Degree of Target-Attacker-Defender Game With Non-zero Capture Radius;Visual Inertial 3D Reconstruction System Based on KLT Optical Flow and Voxel Hash;analysis of Students’ Class Status Based on Deep Learning;fault-Tolerant Attitude control of Hypersonic Flight Vehicles Based on Extended Adaptive Iterative Learning;predefined-Time control of Robot systems with Prescribed performance Guarantees;dynamic Coverage of Unicycle Robots with Anisotropic Sensing for Time-Priority;method for Evaluating the Multi-factor Decision-Making Efficiency Based on Dynamic Bayesian network;submarine Cable Target Search Based on Improved Biological Neural network;end-to-End mmWave-Based Human Pose Estimation from Raw Signal;urban Sewage Treatment Plant Simulation System Based on Industrial internet;Load Matching Design and Multi-objective Optimization of Helicopter EHA;Testability Design and Verification of Typical Functional Links of CNI System;adaptive Sliding Mode Tracking control for Multiple High-Speed Trains with Actuator Saturation and Parameter Uncertainties;Abnormal Data Detection Based on Dual-Factor Weighted SVDD for Multimode Batch Processes;dual-Branch Attention Dense Residual network Based on Negative Image for Image Denoising;Research on Testability Work Planning Technology for Aviation Equipment Based on DoDAF;aero Engine Instability Prediction and Detection Method Based on Gated Recurrent Neural networks;Seven-Degrees-of-Freedom Robotic Arm Path Planning Based on Improved RRT;dynamic Event-Triggered Based Coherent Formation control for Multi-robot systems;adaptive Prescribed Time Prescribed performancecontrol for High-Order No
Recently, with the advances in technologies and the integration of internet of Things (IoT), many fields face difficulties and challenges in implementing IoT based systems and applications. One of the most important d...
详细信息
Accurately measuring Round-Trip Times (RTT) for internet communications is important for various research topics, ranging from protocol performance and congestion control to routing and network security. Unix systems,...
详细信息
The IEEE 802.15.4 standard is designed for low-rate wireless personal area networks (LR-WPANs). Deterministic and Synchronous Multi-channel Extension (DSME) is one of the key Medium Access control (MAC) modes of the I...
详细信息
暂无评论