This book constitutes the refereed proceedings of the 4th International Conference on Cognitive Computation and Systems, ICCCS 2025, held in Chengdu, China, November 2223, 2025.

The 79 full papers presented in this book were carefully selected and reviewed from 199 submissions. The papers are organized in the following topical sections:

Part I: Cognitive computing and Multimodal information processing

Part II: Intelligent robotics

Inhalt

.- MWAA-Former: A Multi-Scale Window Agent Attention Network for Sparse Feature Recognition.
.- VSFNet: A Multi-Scale Feature Enhancement Network Based on Spatial State Modeling and Frequency-Domain Filtering.
.- Frequency-Domain Augmented Re-parameterized YOLO for Medical Object Detection.
.- High-Resolution Class Activation Mapping for Weakly Supervised Object Localization.
.- RDC-XLSTM: A smart contract vulnerability detection with retrieval-enhanced generation and contextual long memory.
.- Co-Support Few-Shot Segmentation via Internal Complement and Intra Balance.
.- A Handheld Multi-Line 3D Scanning Simulation Platform based on Optical-Inertial Signal Generation.
.- Deep neural network based unsupervised artifact detection method on electroencephalography.
.- EEG Lead Selection and Interpretable Deep Learning Model for Depression Diagnosis.
.- CBFOD-YOLO: A Coal Mine Conveyor Belt Foreign Object Detection Model Based on an Improved YOLOv11 Architecture.
.- Performance Analysis of Pancreas Segmentation from CT Images Using Deep Learning.
.- NLW-YOLOv8n: a novel lightweight PCB small target defect detection algorithm.
.- Research of Weld Seam Segmentation Method Based on Point Cloud Data.
.- Gypsum Board Defect Detection Based on Lightweight Networks.
.- Automatic Fabric Defect Detection System: Optimization of Lightweight Architecture with Cloud-Edge Computing Integration.
.- MBCTN: A Multi-Band Convolutional Tensor Network for Speech Imagery EEG Decoding.
.- Semantic-fused Diffusion Model for Thangka Image Super-Resolution Based on Constructed Dataset.
.- TCLo: Transformer optimized for capturing local features in point cloud semantic segmentation.
.- DC-NetRobust Point Cloud Normal Estimation with Dynamic Convolution.
.- Efficient Real-Time Traffic Sign Detection for Autonomous Driving in Adverse Weather Using Deep Learning Models.
.- Energy-Efficient Multi-Layer Adaptive Transmission with Entropy-Weighted TOPSIS for Diverse Contents.
.- Learning to Enhance Low-Light Images via Salient Feature Representation and Feature Integration.
.- Mold positioning and measurement path planning system based on binocular structured light.
.- Improvement of YOLO-v11 Small Target Detection Model Based on Multi-Scale Feature Enhancement and Cross-Scale Attention.
.- Early Fault Monitoring of Engineering Equipment Based on Improved Transformer.
.- Model-independent Debiasing Algorithm Based on Probabilistic Correlation for Scene Graph Generation.
.- Toward Enhanced Neural Decoding: A Framework for Reconstructing EEG Features from fMRI BOLD Signals.
.- A Lightweight Semantic Segmentation Network for UAV Remote Sensing Images: HWD-Deeplab v3+.
.- SLG-Flow: A Unified Framework from Semantic Labeling to Visual Generation.
.- A CNN-based Deep Reinforcement Learning Approach for Imbalanced Walking Direction Recognition.
.- CNN-BiLSTM Hybrid Neural Network Deep Learning Model for Flight Pilot Stress Detection.
.- SemGNG-CPC: A Semantic Topology-Aware Self-Supervised Framework for EEG-Based Emotion Recognition.
.- Medical super-resolution reconstruction network for downstream tasks based on wavelet analysis.
.- An Object Detection and Force Estimation Method based on a Single-point Proximity-capacitance Sensor.
.- Application of large models in the diagnosis of lung diseases.
.- Riesz Transform-Based Fine-Grained Edge Enhancement in Multi-Channel CNNs.
.- Deep Learning-based Inkjet Code Recognition Technology for Cotton Bales.
.- Research on Laser-Based Visual Navigation for AGVs with Multi-Sensor Fusion and Deep Learning.