Computer GraphicsとComputer Visionについて

ECCV 2018気になる論文

ECCV 2018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。


  • MVSNet: Depth Inference for Unstructured Multi-view Stereo
  • PlaneMatch: Patch Coplanarity Prediction for Robust RGB-D Registrati
  • Active Stereo Net: End-to-End Self-Supervised Learning for Active Stereo Systems
  • GAL: Geometric Adversarial Loss for Single-View 3D-Object Reconstruction
  • DeepTAM: Deep Tracking and Mapping


  • Semi-Dense 3D Reconstruction with a Stereo Event Camera
  • Geometric Constrained Joint Lane Segmentation and Lane Boundary Detection
  • Good Line Cutting: towards Accurate Pose Tracking of Line-assisted VO/VSLAM
  • Fully-Convolutional Point Networks for Large-Scale Point Clouds
  • Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement
  • Linear RGB-D SLAM for Planar Environments
  • Estimating Depth from RGB and Sparse Sensing
  • Into the Twilight Zone: Depth Estimation using Joint Structure-Stereo Optimization
  • Layer-structured 3D Scene Inference via View Synthesis
  • ArticulatedFusion: Real-time Reconstruction of Motion, Geometry and Segmentation Using a Single Depth Camera
  • Monocular Scene Parsing and Reconstruction using 3D Holistic Scene Grammar
  • Recovering 3D Planes from a Single Image via Convolutional Neural Networks
  • 3DMV: Joint 3D-Multi-View Prediction for 3D Semantic Scene Segmentation
  • Learning Priors for Semantic 3D Reconstruction
  • Large Scale Urban Scene Modeling from MVS Meshes
  • Learning Category-Specific Mesh Reconstruction from Image Collections
  • StereoNet: Guided Hierarchical Refinement for Real-Time Edge-Aware Depth Prediction
  • Semantically Aware Urban 3D Reconstruction with Plane-Based Regularization
  • BiSeNet: Bilateral Segmentation Network for Real-time Semantic Segmentation
  • RIDI: Robust IMU Double Integration
  • Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network
  • CornerNet: Detecting Objects as Paired Keypoints
  • A Unified Framework for Single-View 3D Reconstruction with Limited Pose Supervision
  • A Minimal Closed-Form Solution for Multi-Perspective Pose Estimation using Points and Lines
  • Learn-to-Score: Efficient 3D Scene Exploration by Predicting View Utility
  • 3D Ego-Pose Estimation via Imitation Learning
  • Look Deeper into Depth: Monocular Depth Estimation with Semantic Booster and Attention-Driven Loss