Your search

Facial expression recognition is a key topic in computer vision, playing a crucial role in non-verbal communication. With the rapid development of artificial intelligence, significant progress has been made in improving recognition accuracy and generalization abilities. Traditional methods often suffer from low precision and poor generalization, while deep learning models have substantially advanced the field. However, deploying deep and complex models on different platforms remains challenging due to their high computational demands and frameworks. Hence, developing an efficient, real-time, and lightweight facial expression recognition system is critical. This study focuses on creating an efficient, accurate, and lightweight real-time facial expression recognition system with an emphasis on cross-platform deployment. It integrates various deep learning and optimization techniques to demonstrate flexibility across platforms. In this context, this study evaluates the performance of 10 advanced CNN models (VGG16, VGG19, ResNet50, etc.) on the facial expression dataset FER2013. YOLOv8 combined with ResNet50 achieved 70.56% accuracy on FER2013, outperforming YOLOv8 alone by 2.1%. Multi-module fusion models (MobileFaceNet, IR50, HyViT, SE) achieved an accuracy of 92.58% and 74.8% on the RAF-DB and FER2013 datasets, respectively, showing superior performance in ablation experiments. Given the significant impact of data quality on model performance, this study performed data cleaning on the FER2013 dataset, resulting in a 3.25% accuracy improvement for the YOLOv8 + ResNet50 model, which reached 73.81%. The high-resolution RAF-DB dataset, with fewer errors, led to improved performance, achieving 92.56% accuracy with the fusion model. A multi-purpose facial expression recognition system, VISTA, was developed using Python and PyQt5. The system supports multiple data formats and provides real-time emotional feedback, thus enhancing its usability for both research and practical applications. Furthermore, the fusion model was quantized using the OpenVINO toolkit, reducing its parameters by 75% while maintaining an accuracy of 91.17%. Inference speed was improved, and XGrad-CAM was employed to enhance model interpretability, revealing that the YOLOv8 + ResNet50 combination more effectively captured facial features. Finally, the high-performance model was successfully deployed on Intel CPUs, NVIDIA GPUs, and embedded devices Raspberry Pi 4B, demonstrating the portability and flexibility of the VISTA system across various platforms. This research provides promising solutions for applications in human-computer interaction, affective computing, and real-time emotional analysis, with significant advancements made in improving system real-time performance, accuracy, and cross-platform deployment capabilities. It contributes to the development of facial expression recognition technology and lays the foundation for its widespread future applications in fields such as smart healthcare, business analytics, education, and mental health.

This doctoral research delves into the transformative potential of Hyperledger Fabric Blockchain and the Internet of Things (IoT) within business management, specifically in the development and implementation of a Computerised Maintenance Management System (CMMS). It suggests that merging these advanced technologies could revolutionise maintenance management and overall system performance. The study assesses the impact on fundamental business processes within the IoT paradigm, highlighting the role of the Hyperledger Fabric blockchain network in ensuring data integrity and enhancing transparency. The integration of Blockchain protocol with IoT offers efficient data transactions, thereby improving business data management and decision-making. The research further validates the robustness of Fabric release V2.4 for CMMS development. The study concludes by emphasising the need for additional research to understand long-term implications and challenges in different business environments

This thesis introduces, implements and evaluates an innovative concept for assessing driving behavior in public transportation through Mobile Crowd Sensing (MCS), under the field of Advanced Public Transportation System (APTS) - a sub-group of Intelligent Transportation Systems (ITS). Aggressive driving behavior is known to be a cause of avoidable accidents and to increase fuel consumption. In public transportations, it is also a case for costumers’ dissatisfaction. Monitoring the quality of driving behavior is a key element to overcome this issue and to improve road safety and customer satisfaction. In this research project, a software application (app) for mobile devices was developed as an experimental tool / proof-of-concept, to monitor aggressive driving behavior in bus drivers, collecting data coming from mobile device’s accelerometer and passengers’ qualitative evaluation. The experimental procedure took place in public transportation in Macau (bus only) and consisted of data collection of drivers’ aggressive driving behavior using the developed application. The analysis of collected data suggests that MCS is a viable way to assess drivers’ behavior in public transportation, thus contributing to the improvement of the service and increase of road safety. Although the methodology has been tailor-made for Macau public transportation, it is believed that the same concept can be applied to other cities, leading them towards the goal of becoming smarter cities. Keywords: driving behavior; mobile crowd sensing; crowdsourcing; smart city; advanced public transportation system; intelligent transportation system; road safety; mobile device accelerometer

This thesis articulates the development of a holistic approach to enhance learning and teaching in an object-oriented programming course. Starting with the premise that it is not possible to improve teaching without understanding how students learn programming, this thesis embodies the processes and reflections experienced while applying knowledge of how students learn programming, to design a learning environment that enhances learning outcomes. First, a theoretically based framework for the teaching of the course is developed. A holistic approach using a plurality of pedagogic theories, taxonomies, and instructional designs is employed to bridge the gaps between the bodies of knowledge relating to the ways that students approach programming and the application of this knowledge to design the course. Second, in two cycles of action research, the course is implemented and the analysis of its outcome is conducted using mixed methods data collection techniques. The evaluation is integrative and seeks multiple forms of evidence for student engagement and improved learning. The original contributions from this research in the form of new initiatives, perceptions, and understandings, as well as implications for theory and practice are described. A claim to knowledge is established by explaining the significance of the research to student learning, personal practice and beliefs, institutional influence, and potential for influence on computing education research. Quality criteria are applied to assess the validity and rigor of the action research project, and the research is appraised as a scholarly enquiry and a transformative process that led to innovative forms of thinking and acting