A5044164299- Emotion and Mood Recognition
- Video Surveillance and Tracking Methods
- Advanced Neural Network Applications
- Autonomous Vehicle Technology and Safety
- Face and Expression Recognition
- Face recognition and analysis
- Image Retrieval and Classification Techniques
- Image Processing and 3D Reconstruction
- Generative Adversarial Networks and Image Synthesis
- Gaze Tracking and Assistive Technology
- Remote Sensing and LiDAR Applications
- Medical Imaging and Analysis
- EEG and Brain-Computer Interfaces
- Artificial Intelligence in Healthcare
- Machine Learning in Healthcare
- Domain Adaptation and Few-Shot Learning
- Human Pose and Action Recognition
- Dental Radiography and Imaging
- Speech and Audio Processing
- Pain Mechanisms and Treatments
- Sleep and Work-Related Fatigue
- Infrared Thermography in Medicine
- Pain Management and Placebo Effect
Imperial College London
2012-2016
Karlsruhe University of Education
2009
Pain-related emotions are a major barrier to effective self rehabilitation in chronic pain. Automated coaching systems capable of detecting these potential solution. This paper lays the foundation for development such by making three contributions. First, through literature reviews, an overview how pain is expressed and motivation it physical provided. Second, fully labelled multimodal dataset (named `EmoPain') containing high resolution multiple-view face videos, head mounted room audio...
This paper is about estimating intensity levels of Facial Action Units (FAUs) in videos as an important step toward interpreting facial expressions. As input features, we use locations landmark points detected video frames. To address uncertainty input, formulate a generative latent tree (LT) model, its inference, and novel algorithms for efficient learning both LT parameters structure. Our structure iteratively builds by adding either new edge or hidden node to LT, starting from initially...
Certain inner feelings and physiological states like pain are subjective that cannot be directly measured, but can estimated from spontaneous facial expressions. Since they typically characterized by subtle movements of parts, analysis the details is required. To this end, we formulate a new regression method for continuous estimation intensity behavior interpretation, called Doubly Sparse Relevance Vector Machine (DSRVM). DSRVM enforces double sparsity jointly selecting most relevant...
Human nonverbal behavior recognition from multiple cues and modalities has attracted a lot of interest in recent years. Despite the interest, many research questions, including type feature representation, choice static vs. dynamic classification schemes, number or to use, optimal way fusing these, remain open questions. This paper compares frame-based vs window-based representation employs schemes for two distinct problems field automatic human analysis: multicue discrimination between...
Standard machine learning approaches require centralizing the users' data in one computer or a shared database, which raises privacy and confidentiality concerns. Therefore, limiting central access is important, especially healthcare settings, where regulations are strict. A potential approach to tackling this Federated Learning (FL), enables multiple parties collaboratively learn prediction model by using parameters of locally trained models while keeping raw training locally. In context...
Users react differently to non-relevant and relevant tags associated with content. These spontaneous reactions can be used for labeling large multimedia databases. We present a method assess tag relevance images using the non-verbal bodily responses, namely, electroencephalogram (EEG), facial expressions, eye gaze. conducted experiments in which 28 were shown subjects once correct another time incorrect tags. The goal of our system is detect responses consequently filter them out. Therefore,...
Images are composed as a hierarchy of object parts. We use this insight to create generative graphical model that defines hierarchical distribution over image Typically, leads intractable inference due loops in the graph. propose an alternative structure, Dense Latent Tree (DLT), which avoids and allows for efficient exact inference, while maintaining dense connectivity between parts hierarchy. The usefulness DLTs is shown example task completion on partially observed MNIST Fashion-MNIST...
Autonomous vehicles require knowledge of the surrounding road layout, which can be predicted by state-of-the-art CNNs. This work addresses current lack data for determining lane instances, are needed various driving manoeuvres. The main issue is time-consuming manual labelling process, typically applied per image. We notice that car itself a form annotation. Therefore, we propose semi-automated method allows efficient image sequences utilising an estimated plane in 3D based on where has...