CIFAR10-DVS Object Classification Event Stream Dataset

CIFAR stands for Canadian Institutes for Advanced Research, DVS stands for Dynamic Vision Sensor, and CIFAR10-DVS is an event stream dataset for object classification. The dataset has 10 categories (airplanes, cars, birds, cats, deer, dogs, frogs, horses, ships, and trucks), with a total of 10,000 event streams, and is of medium difficulty. These event streams are converted from frame-based images in the CIFAR-10 dataset through event-based image sensors, using image repetitive closed-loop smoothing (RCLS) motion to achieve the conversion. Due to the conversion, the image produces rich local intensity changes in continuous time, which are quantified by the event camera on a pixel basis.