Efficient, Data-Driven Perception with Event Cameras (Ph.D. Defense of Daniel Gehrig)

Event cameras are taking the world by storm because they promise efficient, low-latency, and robust perception, especially in challenging scenarios where standard cameras suffer from blur, saturation, or high latencies. However, event cameras only measure their sparse and asynchronous intensity changes, not images, and thus, new algorithms need to be developed. The best event-based algorithms nowadays convert events into frames and then process them with deep neural networks. While these methods are highly robust, they are also highly inefficient. The Ph.D. thesis of Daniel Gehrig investigates ways in which to reconcile the efficiency and robustness of event-based perception and optimally fuse it with complementary image sensors. His thesis contributes to efficient and asynchronous object detection with events and frames, which opens the door to robust and low-latency event-based vision. Reference: Daniel Gehrig Efficient, Data-Driven Perception with Event Cameras Google Scholar: https://scholar.google.com/citations?... Our research page on event-based vision: http://rpg.ifi.uzh.ch/research_dvs.html For event-camera datasets, see here: 1. https://dsec.ifi.uzh.ch/ 2. http://rpg.ifi.uzh.ch/davis_data.html 3. https://github.com/uzh-rpg/event-base... For an event camera simulator: http://rpg.ifi.uzh.ch/esim For a survey paper on event cameras, see here: https://rpg.ifi.uzh.ch/docs/EventVisi... Other resources on event cameras (publications, software, drivers, where to buy, etc.): https://github.com/uzh-rpg/event-base... Affiliation: Daniel Gehrig is with the Robotics and Perception Group, Dept. of Informatics, University of Zurich, and Dept. of Neuroinformatics, University of Zurich and ETH Zurich, Switzerland http://rpg.ifi.uzh.ch/