How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
Citations since 2017
3 Research Items
Passionate about programming since a young age, I want to push the limits of Artificial Intelligence with an emphasis on the last word through few-shot learning in Computer Vision applications. My PhD thesis will be on the application of few-shot learning to hand and object pose estimation during manipulation, for object-specific adaptation of neural networks. "What I cannot create, I do not understand" - Richard Feynman
Understanding hand-object pose with computer vision opens the door to new applications in mixed reality, assisted living or human-robot interaction. Most methods are trained and evaluated on balanced datasets. This is of limited use in real-world applications; how do these methods perform in the wild on unknown objects? We propose a novel benchmark...
The lack of assistive Sign Language technologies for members of the Deaf community has impeded their access to public information, and curtailed their civil rights and social inclusion. In this paper, we introduce a novel proof-of-concept method for end-to-end Sign Language to speech translation without an intermediate text representation.We propos...
Drone racing is a recreational sport in which the goal is to pass through a sequence of gates in a minimum amount of time while avoiding collisions. In autonomous drone racing, one must accomplish this task by flying fully autonomously in an unknown environment by relying only on computer vision methods for detecting the target gates. Due to the ch...
Current hand tracking for VR/AR interfaces focuses on the manipulation of virtual objects such as buttons, sliders and knobs. Such tracking is most often based on tracking each hand independently and when hands become partially occluded or are grasping a real object the hand tracking often fails. Tracking the hands during the manipulation of real-world objects opens up AR/VR to much richer forms of interaction and would provide the basis for activity recognition and the display of detailed contextual information related to the task at hand. This PhD project involves researching the tracking of unmodified hands with an ego-centric camera (2D and 3D) in the presence of partial occlusions. Technologies will include the use of deep learning models in combination with 3D models to determine hand pose in the presence of occlusion. Our approach will also exploit high level knowledge about object affordances and common hand grasp configurations which is commonly used in Robotic grasping.