Event Frame
One method of representing event camera data that accumulates events within a time window into a two-dimensional image with two channels for positive and negative polarity. It can be fed directly into CNNs but loses the temporal ordering of events.
Timeline (last 90 days)
Event frame representation accumulates events within a time window into a two-dimensional image, usually using two channels to represent positive and negative polarity respectively; it is closest to traditional frames and can be fed directly into a CNN.
Event frame representation loses the temporal order of events, and the choice of window size has a large impact.
Event frames suit object detection in static scenes, time surfaces suit motion estimation, and voxel grids are used for tasks requiring fine spatiotemporal information.
The tutorial author recommends starting with simple event frames and gradually trying more complex representations.
All Facts (4)
Event frame representation accumulates events within a time window into a two-dimensional image, usually using two channels to represent positive and negative polarity respectively; it is closest to traditional frames and can be fed directly into a CNN.
50%UnverifiedEvent frame representation loses the temporal order of events, and the choice of window size has a large impact.
50%UnverifiedEvent frames suit object detection in static scenes, time surfaces suit motion estimation, and voxel grids are used for tasks requiring fine spatiotemporal information.
50%UnverifiedThe tutorial author recommends starting with simple event frames and gradually trying more complex representations.
50%