Faranak Shamsafar, Samuel Woerz, Rafia Rahim, Andreas Zell
TL;DR该研究提出了两种轻量级的立体视觉模型,采用 2D 和 3D 卷积进行编解码。研究表明,这两种模型有效地减少了计算开销,同时保持了准确性。
Abstract
Recent methods in stereo matching have continuously improved the accuracy
using deep models. This gain, however, is attained with a high increase in
computation cost, such that the network may not fit even on a m