Katerina Fragkiadaki

Katerina Fragkiadaki is interested in building machines that understand the stories that videos portray, and, inversely, in using videos to teach machines about the world. She is working on teaching machines basic common sense. Her recent works use recurrent nets with 3D representation bottlenecks that disentangle camera motion from appearance, while being optimized end-to-end for a final task, such as view prediction or 3D object detection. In this way, her models learn object permanence, size constancy, occlusions and dis-occlusion relationships, useful for 3D object detection, affordance learning and language grounding.