Stereo matching is a long-standing task in computer vision. It aims to recover the dense correspondences between image pairs to recover their geometry. Recently, with the rise of deep learning, convolutional neural network methods have replaced more traditional stereo matching methods. However, there are still two major problems that remain unsolved: predicting accurate depth boundaries and generating high-resolution outputs with limited memory and computation. The overall goal for this work is to build a stereo matching algorithm that can work at a very high resolution and predict sharp and precise object boundaries. The team have a sensor in their lab that captures at 12Mpx resolution, which they want the algorithm to work with, and an algorithm capable of providing precise 3D geometry was desirable. Fabio Tosi is a postdoc at the University of Bologna and a visiting PhD student at the University of Tübingen and Max Planck Institute for Intelligent Systems where he joined the Autonomous Vision Group. SMD-Nets: Stereo Mixture Density Networks 6 DAILY CVPR Wednesday Presentation Yiyi Liao is a postdoc in the Autonomous Vision Group where she and Fabio work under the supervision of Professor Andreas Geiger. Their paper focuses on stereo matching, and they speak to us ahead of their presentation today.