Hara Kosuke

TL;DR: In this paper, the authors propose EventNet, a neural network designed for real-time processing of asynchronous event streams in a recursive and event-wise manner, which enables processing of 1 mega or more events per second on standard CPU.

TL;DR: In this article, the authors proposed a risk estimation device that includes a feature amount extraction part (41) for extracting a plurality of feature amounts representing features of conditions around a vehicle (10) on the basis of sensor information generated by a sensor (11), and a scene estimation part (42) for acquiring, about each scene, a probability that a condition around the vehicle corresponds to the scene by applying the plurality of features to a probability distribution.

TL;DR: In this paper, a line segment extraction unit (21) is used to extract a first line segment from a first image created at a first time by a first imaging unit (2-2) mounted on a vehicle (10); a projection unit (22) for identifying a first area in a real space corresponding to a matching area including the first line segments in the first image satisfying predetermined conditions for specifying the relationship of the vehicle with structures around the vehicle; and an area in an second image corresponding to the matching area by projecting the first area on the second image created by a second

TL;DR: In this article, an arithmetic device capable of calculating a correlation of a function which takes a discrete value at a high speed was proposed. But it was not shown how to calculate the correlation between the template image and query image.