Cognitive Systems and Information Processing: 8th International Conference, ICCSIP 2023, Luoyang, China, August 10–12, 2023, Revised Selected Papers, ... in Computer and Information Science)

The two-volume set CCIS 1918 and 1919 constitutes the refereed post-conference proceedings of the 8th International Conference on Cognitive Systems and Information Processing, ICCSIP 2023, held in Luoyang, China, during August 10–12, 2023. The 52 full papers presented in these proceedings were carefully reviewed and selected from 136 submissions. The papers are organized in the following topical sections: Volume I : Award; Algorithm & Control; and Application. Volume II: Robotics & Bioinformatics; and Vision. Preface Organization Contents – Part II Contents – Part I Robotics and Bioinformatics Iris Recognition Network Based on Full Coordinate Attention 1 Introduction 2 Methodology 2.1 Iris Recognition Process 2.2 Iris Pre-Processing 2.3 Attention Blocks 3 Experiments 3.1 Datasets and Metrics 3.2 Experiment Setup 3.3 Ablation Studies 4 Conclusions References Obstacle Avoidance Control Method for Robotic Assembly Process Based on Lagrange PPO 1 Introduction 2 Related Work 3 Methodology 3.1 Constrained Markov Decision Process 3.2 Safe RL 4 Experimental 4.1 Experimental Setup 4.2 Experimental Analysis 5 Conclusion and Prospect References Zero-Shot Sim-To-Real Transfer of Robust and Generic Quadrotor Controller by Deep Reinforcement Learning 1 Introduction 2 Problem Statement 3 Training Environment Implementation 3.1 Rigid Body Dynamics of Quadrotor 3.2 Motor Model in Simulator 3.3 Environment Framework for Real Flight Test 3.4 Training Environment Framework Design 4 Control Policy Training Considering Sim-To-Real Transfer 4.1 Network of Policy 4.2 Initial State and Target State 4.3 RL Cost Function 4.4 Sim-To-Real Transfer 5 Real-World Experiments 5.1 Real Flight Test on Different Quadrotors 5.2 Real Flight Robustness Test 6 Conclusions References Robot Manipulation Strategy for Explosives Disposal in Cluttered Scenes 1 Introduction 2 Related Work 2.1 Object Detection 2.2 Object Relationship Reasoning 2.3 Robot Manipulation Strategies 3 Proposed Framework 3.1 Object Detection 3.2 Object Relationship Reasoning 3.3 Manipulation Control 3.4 Train 4 Experiments 4.1 Indoor Environmental Experiments 4.2 Outdoor Environmental Experiments 5 Conclusion References Adaptive Neural Composite Learning Finite-Time Control for a QUAV with Guaranteed Tracking Performance 1 Introduction 2 Problem Statements and Preliminaries 2.1 Model Description 2.2 Finite-Time Prescribed Performance Function 2.3 RBFNNs 3 Main Result 3.1 Adaptive Composite Learning Finite-Time Controller Design 3.2 Stability Analysis 4 Simulation Result 5 Conclusion References Multimodal Wearable Device Signal Based Epilepsy Detection with Multi-scale Convolutional Neural Network 1 Introduction 2 Methodology 2.1 Data Acquisition 2.2 Proposed MSCNN Based Seizure Detection 2.3 Parameter Setting 3 Experimental Result and Analysis 3.1 Experimental Evaluation 3.2 Comparison with Mainstream Methods 3.3 Ablation Study 3.4 Epilepsy Detection Results 4 Conclusions References Improving EEG-Based Continuous Grip Force Decoding in Grasp-Lift Tasks by Considering Grip Force Levels 1 Introduction 2 Materials and Method 2.1 WAY-EEG-GAL Dataset 2.2 Grip Force Level Analysis 2.3 Data Preprocessing 2.4 Decode Continuous Grip Force 2.5 Pseudo-Online Experiment 3 Result 4 Conclusion and Discussion References 3D Path Planning and Tracking of Quadrotors Based on Improved Whale Optimization Algorithm 1 Introduction 2 Standard Whale Optimization Algorithm (WOA) 2.1 Hunting Prey Stage 2.2 Surrounding Prey Stage 2.3 Bubble Net Predation Stage 3 Improved Whale Optimization Algorithm (SAMWOA) 3.1 Singer Chaotic Mapping 3.2 Adaptive Cognitive Factor 3.3 Wavelet Mutation Mechanism 3.4 Cubic Spline Interpolation 4 Simulation Results and Analysis 4.1 Environment Map Modeling 4.2 3D Path Planning Experiment 4.3 3D Path Tracking Experiment 5 Conclusion References Multimodal Pneumatic Control System for Soft Robot 1 Introduction 2 Method and Martials 2.1 System Design 2.2 Multi-mode Pressure Control Method 3 Experiments and Discussion 3.1 Prototype Production and Performance Experiment 3.2 Application Verification of Variable Stiffness Four-Chamber Soft Robot 3.3 Application Verification of Six-Finger Soft Gripper 4 Conclusion References Enhancing Robot Manipulation Skill Learning with Multi-task Capability Based on Transformer and Token Reduction 1 Introduction 2 Related Work 3 Problem Definition 4 Method 4.1 Background 4.2 VL Fusion 4.3 CSALearner 4.4 Transformer 4.5 Action Prediction 4.6 Training and Inference 5 Experiments 6 Conclusion References Vision Exploring Model Depth Adaptation in Image Super-Resolution for Efficient Inference 1 Introduction 2 Adaptive Model Depth Adjustment 2.1 MDASR Network 2.2 Depth Adaptive Residual Blocks 2.3 Training Loss 3 Experiments 3.1 Settings 4 Conclusion References RGB to L * a * b * Color Prediction Model Based on Color Cards 1 Introduction 2 Equipment and Materials 3 Model Construction 3.1 Linear Models 3.2 Quadratic Model 3.3 Neural Network Model 3.4 Extreme Learning Machine 4 Experimental Results and Analysis 4.1 Evaluation Indexes 4.2 Data Acquisition and Pre-processing 4.3 Comparison of Model Effects 5 Conclusion References A Diverse Environment Coal Gangue Image Segmentation Model Combining Improved U-Net and Semi-supervised Automatic Annotation 1 Introduction 2 Related Work 3 Deep Learning Semantic Segmentation Model 4 Realization Methods of Human-in-the-Loop and Active Sensing 4.1 Pre-training of the Human-in-the-Loop Network 4.2 Automatic Annotation and Filtering 4.3 Correct and Update the Data 5 Test Results and Analysis 5.1 Performance Evaluation Index of the Model 5.2 Training and Validation of Different Segmentation Models 5.3 Model Improvement for Uneven Illumination and Fine Coal Gangue 5.4 Setting of Pixel Ratio Index 6 Conclusion References RB-LIO: A SLAM Solution Applied to Large-Scale Dynamic Scenes with Multiple Moving Objects 1 Introduction 2 Related Work 3 RB-LIO Approach 3.1 System Overview 3.2 Dynamic Object Segmentation 3.3 Back-End Loop Closure Optimization 4 Experiments 4.1 Experiments on KITTI and MulRan Datasets 4.2 Campus Dataset Experiment 5 Conclusion References Object Detection with Depth Information in Road Scenes 1 Introduction 2 Related Works 2.1 Object Detection 2.2 Depth Estimation 3 Research Methods 3.1 Loss Function 3.2 Depth Estimation Sub-Network 3.3 Object Detection Sub-Network 3.4 Shared Units 4 Results 4.1 Experiment Environments 4.2 Result Analysis 5 Conclusion References Mobilenetv2-SSD Target Detection Method Based on Multi-scale Feature Fusion 1 Introduction 2 Related Work 3 Mobilenetv2 Based SSD Target Detection Algorithm 3.1 MobileNetV2 Network 3.2 MobileNetV2-SSD 4 Lightweight Model Optimization Based on Feature Fusion 5 Experiments and Results 6 Conclusion References No-Reference Point Cloud Quality Assessment via Contextual Point-Wise Deep Learning Network 1 Introduction 2 Related Work 2.1 FR and RR Methods 2.2 NR Methods 3 Methodology 3.1 Overview 3.2 Segmentation Sub-Patches 3.3 Offset Attention Feature Encoder 3.4 Quality Score Prediction 4 Experiments 4.1 Experiment Implementation and Criteria 4.2 Prediction Performance on SJTU Database 4.3 Prediction Performance on WPC Database 4.4 Ablation Study 4.5 Visualization Comparisons 5 Conclusion References Measurement of Solder Wetting Angle Based on Image Processing and Polynomial Fitting 1 Introduction 2 Methods and Experiments 2.1 Acquisition of Solder Images 2.2 Solder Contour Extraction 2.3 Calculation of the Wetting Angle of Solder 3 Results and Discussion 4 Conclusion References Infrared Aircraft Anti-interference Recognition Based on Feature Enhancement CenterNet 1 Introduction 2 CenterNet 3 CenterNet Infrared Aircraft Anti interference Recognition Model Based on Feature Enhancement 3.1 Feature Fusion Based on Channel Feature Enhancement 3.2 Attention Mechanism for CenterNet 4 Experiment and Analysis 4.1 DataSet 4.2 Performance Evaluation Index 4.3 Ablation Experiment 4.4 Comparison of Different Algorithms 5 Conclusion References MHRN-ST: Go Board Recognition With Multi-stage Highlight Removal Network Based On Swin Transformer 1 Introduction 2 Related work 2.1 Methods Based on Traditional Image Processing 2.2 Method Based on Deep Learning 3 Methodology 3.1 Overview 3.2 Specular Highlight Detection and Go Board Recognition 3.3 Specular Highlight Removal 3.4 Loss Functions 3.5 Implementation Details 4 Experimental Results 4.1 Test Results 4.2 Comparisons 4.3 Ablation Studies 4.4 Limitations 5 Conclusion References Construction of Minecraft Virtual Reality Scene Based on Voxel Point Cloud 1 Introduction 2 Related Works 3 Methodology 3.1 Point Cloud Preprocessing 3.2 Voxel Point Cloud Segmenting and Coloring 3.3 Minecraft Based Voxelized Scene Exporting 4 Experiment 4.1 Experiment Conditions 4.2 Results and Analysis 5 Conclusion References Author Index