Section: Highlights of the Year

Highlights of the Year

Vanishing point detection is and old problem of computer vision. We introduced this year a new method based on the a contrario methodology to solve this problem. By fractioning the 2-D search of meaningful vanishing points into three 1-D searches of meaningful events (Zenith line, Horizon line, and Vanishing points), we not only achieve state-of-the-art performance w.r.t. computation times and accuracy of the horizon line, but also yields more relevant vanishing points than the previous top-ranked methods. This work was presented at ECCV 2018 [23] and the associated code is freely distributed.