Guest Editor(s)
Assoc. Prof. Christine Dewi
Email: christine.dewi@uksw.edu
Affiliation: Faculty of Information Technology, Satya Wacana Christian University, Salatiga, Indonesia
Homepage:
Research Interests: image processing, computer vision, object detection and recognition, artificial intelligence, knowledge-guided, machine learning
Prof. Rung-Ching Chen
Email: crching@cyut.edu.tw
Affiliation: Department of Information Management, Chaoyang University of Technology, Taichung, Taiwan
Homepage:
Research Interests: pattern recognition and knowledge engineering, iot and data analysis, applications of artificial intelligence, computer vision, image processing
Dr. Muhammad Asad Arshed
Email: asad.arshed@umt.edu.pk
Affiliation: School of Systems and Technology, University of Management and Technology, Lahore, Pakistan
Homepage:
Research Interests: machine learning, deep learning, pattern recognition, medical imaging, computer vision, bioinformatics, natural language processing
Summary
Recent advances in deep learning have significantly improved object detection and image classification; however, many models remain fragile under real-world conditions such as domain shift, class imbalance, adverse weather, and sensor noise. Their black-box nature further limits transparency and trust in safety- and mission-critical applications. Addressing these limitations requires moving beyond purely data-driven approaches toward knowledge-enhanced vision systems.
This Special Issue focuses on integrating explicit structured knowledge—such as knowledge graphs, ontologies, label hierarchies, physical or causal priors, and expert-defined constraints—into detection and classification pipelines to improve robustness, interpretability, and generalization. We invite methodological, benchmarking, and application-driven studies across domains including environmental monitoring, medical imaging, remote sensing, robotics, intelligent transportation, and industrial inspection. Submissions should demonstrate reproducible evaluation and clear evidence that knowledge integration leads to improved reliability and explainability under real-world conditions.
Topics of interest include, but are not limited to:
· Knowledge-enhanced object detection and image classification using explicit structured knowledge (e.g., knowledge graphs, ontologies, label hierarchies, scene graphs).
· Graph-based vision models (e.g., GNN, GCN, GAT) for modelling label dependencies, object relationships, and contextual reasoning.
· Knowledge-guided training objectives and regularization strategies, including logic-based constraints, hierarchy-aware learning, and semantic consistency.
· Interpretable, explainable, and uncertainty-aware detection and classification for trustworthy and safety-critical deployment.
· Robust recognition under real-world challenges, including domain shift, long-tailed class distributions, and class imbalance.
· Multi-label and multi-task learning frameworks that integrate structured priors and relational reasoning.
· Domain adaptation, and transfer learning with explicit knowledge integration.
· Benchmarking, dataset development, and reproducibility studies for robust and interpretable computer vision systems.
Keywords
knowledge-enhanced object detection and image classification, graph-based vision models, knowledge-guided training objectives and regularization strategies, interpretable explainable and uncertainty-aware detection and classification, robust recognition under real-world challenges, multi-label and multi-task learning with structured priors, domain adaptation and transfer learning with explicit knowledge integration, benchmarking dataset development and reproducibility for robust interpretable computer vision
Login
Register
Submit a Paper
Propose a Special lssue