Existing object tracking algorithms generally use some form of local optimisation, assuming that an object's position and shape change smoothly over time. In some situations ...
Abstract. This paper presents an approach on high-level feature detection using a region thesaurus. MPEG-7 features are locally extracted from segmented regions and for a large set...
Deep Belief Networks (DBNs) are hierarchical generative models which have been used successfully to model high dimensional visual data. However, they are not robust to common vari...
For many biomedical modelling tasks a number of different types of data may influence predictions made by the model. An established approach to pursuing supervised learning with ...
Yiming Ying, Colin Campbell, Theodoros Damoulas, M...
This paper introduces our head pose estimation system that localizes nose-tip of the faces and estimates head poses in studio quality pictures. After the nose-tip in the training d...