Computer GraphicsとComputer Visionについて

ICCV 2017気になる論文

ICCV 2017で発表される気になる論文です。主に3次元再構成系です。

  • Probabilistic Structure from Motion with Objects (PSfMO)
  • Efficient Global 2D-3D Matching for Camera Localizationin a Large-Scale 3D Map
  • 3D Surface Detail Enhancement from A Single Normal Map
  • SurfaceNet: An End-to-end 3D Neural Network for Multiview Stereopsis
  • Octree Generating Networks: Efficient Convolutional Architectures for High-resolution 3D Outputs
  • Unsupervised Learning of Stereo Matching
  • Quasiconvex Plane Sweep for Triangulation with Outliers
  • TorontoCity: Seeing the World with a Million Eyes
  • Reconfiguring the Imaging Pipeline for Computer Vision
  • Surface Normals in the Wild
  • Semantic Line Detection and Its Applications
  • Semantically Informed Multiview Surface Refinement
  • Modeling Urban Scenes From Pointclouds
  • From Point Clouds to Mesh using Regression
  • RoomNet: End-to-End Room Layout Estimation
  • What Is Around The Camera?
  • Learning 3D Object Categories by Looking Around Them
  • CAD Priors for Accurate and Flexible Instance Reconstruction
  • SSD-6D: Making RGB-Based 3D Detection and 6D Pose Estimation Great Again
  • Practical Projective Structure from Motion (P2SfM)
  • Distributed Very Large Scale Bundle Adjustment by Global Camera Consensus
  • FLaME: Fast Lightweight Mesh Estimation using Variational Smoothing on Delaunay Graphs