Daily-Paper-CVPR20-in-3D

Conference Home Seattle, Washington

Get more information, please subscribe the account "3D Daily" in Wechat.

Includes 3D Classification/Segmentation, Detection, Generation/Reconstruction, 3D Face, Pose Estimation, Matching, Keypoints, Layout, Depth, Surfaces, Texture, Graph, Dataset, et al.

More details about the papers please see the subject documents for papers' abstraction and pipeline.

-Classification/Segmentation

-Detection

-Generation/Reconstruction

-Pose Estimation

-Keypoints, Layout, Depth, Surface

-Matching, Tecture, Datasets

-3D Face, Graph, Others

1. The first link "PR20" after each title is the paper's address from the arXiv websites.
2. The link "TF/Torch" means the code based on which platform. "Code" means the code has not released yet.
3. The link "Account" introduces the paper in Chinese from the account in Wechat.
4. The link like "Graph" means this paper also in the "Graph" subject. "C/S" means "Classification/Segmentation", et al.

---

CS

1. PointGroup: Dual-Set Point Grouping for 3D Instance Segmentation PR20
2. PointGLR: Global-Local Bidirectional Reasoning for Unsupervised Representation Learning of 3D Point Clouds PR20 Torch
3. 3D-MPA: Multi Proposal Aggregation for 3D Semantic Instance Segmentation PR20 Project Video
4. PolarNet: An Improved Grid Representation for Online LiDAR Point Clouds Semantic Segmentation PR20 Torch
5. DualConvMeshNet: Joint Geodesic and Euclidean Convolutions on 3D Meshes PR20 Torch
6. Learning Unsupervised Hierarchical Part Decomposition of 3D Objects from a Single RGB Image PR20 Torch
7. Multi-Path Region Mining For Weakly Supervised 3D Semantic Segmentation on Point Clouds PR20
8. Learning to Segment 3D Point Clouds in 2D Image Space PR20 TF
9. xMUDA： Cross-Modal Unsupervised Domain Adaptation for 3D Semantic Segmentation PR20 Code
10. FPConv: Learning Local Flattening for Point Convolution PR20 Torch
11. OccuSeg: Occupancy-aware 3D Instance Segmentation PR20 Graph
12. GridGCN: Grid-GCN for Fast and Scalable Point Cloud Learning PR20 Account Graph
13. PointAugment: An Auto-Augmentation Framework for Point Cloud Classification PR20 Code
14. PointASNL: Robust Point Clouds Processing using Nonlocal Neural Networks with Adaptive Sampling PR20 TF
15. RandLANet: Efficient Semantic Segmentation of Large-Scale Point Clouds PR20 TF Account
16. Weakly Supervised Semantic Point Cloud Segmentation:Towards 10X Fewer Labels PR20 Code

---

Detection

1. FocalMix: Semi-Supervised Learning for 3D Medical Image Detection PR20
2. LiDAR-based Online 3D Video Object Detection with Graph-based Message Passing and Spatiotemporal Transformer Attention PR20 Code
3. Learning to Detect 3D Objects and Predict their 3D Shapes PR20
4. PointGNN: Graph Neural Network for 3D Object Detection in a Point Cloud PR20 TF Graph
5. MonoPair: Monocular 3D Object Detection Using Pairwise Spatial Relationships PR20
6. MotionNet: Joint Perception and Motion Prediction for Autonomous Driving Based on Bird's Eye View Maps PR20 Code
7. HVNet: Hybrid Voxel Network for LiDAR Based 3D Object Detection PR20 Account
8. Disp R-CNN: Stereo 3D Object Detection via Shape Prior Guided Instance Disparity Estimation PR20 Torch
9. End-to-End Pseudo-LiDAR for Image-Based 3D Object Detection PR20 Code
10. UCNet: Uncertainty Inspired RGB-D Saliency Detection via Conditional Variational Autoencoders PR20 Oral Code
11. MLCVNet: Multi-Level Context VoteNet for 3D Object Detection PR20 Code
12. Probabilistic Orientated Object Detection in Automotive Radar PR20

---

GenRecons

1. PIFuHD: Multi-Level Pixel-Aligned Implicit Function for High-Resolution 3D Human Digitization PR20 Project
2. Local Implicit Grid Representations for 3D Scenes PR20 Video
3. DeepCap: Monocular Human Performance Capture Using Weak Supervision PR20
4. 3DSketchAwareSSC: 3D Sketch-aware Semantic Scene Completion via Semi-supervised Structure Prior PR20 Project Code
5. PQNet: A Generative Part Seq2Seq Network for 3D Shapes PR20 Torch Account
6. Implicit Functions in Feature Space for 3D Shape Reconstruction and Completion PR20 Code
7. InPerfectShape: Certifiably Optimal 3D Shape Reconstruction from 2D Landmarks PR20
8. PFNet: Point Fractal Network for 3D Point Cloud Completion PR20 Torch Account
9. Cascaded Refinement Network for Point Cloud Completion PR20 Code
10. Learning 3D Semantic Scene Graphs from 3D Indoor Reconstructions PR20 Video
11. ARCH: Animatable Reconstruction of Clothed Humans PR20

---

PoseEstimation

1. Towards Better Generalization: Joint Depth-Pose Learning without PoseNet PR20 Torch Depth
2. HandVoxNet: Deep Voxel-Based Network for 3D Hand Shape and Pose Estimation from a Single Depth Map PR20
3. ONP: Optical Non-Line-of-Sight Physics-based 3D Human Pose Estimation PR20 Project
4. HOPENet: A Graph-based Model for Hand-Object Pose Estimation PR20 Project Torch Account Graph
5. Compressed Volumetric Heatmaps for Multi-Person 3D Pose Estimation PR20 Torch
6. Estimating 6D Pose of Objects with Symmetries PR20 Project
7. BodiesatRest: 3D Human Pose and Shape Estimation from a Pressure Image using Synthetic Data PR20
8. G2LNet: Global to Local Network for Real-time 6D Pose Estimation with Embedding Vector Features PR20 Torch
9. Cross-View Tracking for Multi-Human 3D Pose Estimation at over 100 FPS PR20
10. Weakly-Supervised Mesh-Convolutional Hand Reconstruction in the Wild PR20 Oral Project
11. Lightweight Multi-View 3D Pose Estimation through Camera-Disentangled Representation PR20
12. Self-Supervised 3D Human Pose Estimation via Part Guided Novel Image Synthesis PR20 Oral Project
13. MoreFusion: Multi-object Reasoning for 6D Pose Estimation from Volumetric Fusion PR20
14. Weakly-Supervised 3D Human Pose Learning via Multi-view Images in the Wild PR20
15. Video Inference for Human Body Pose and Shape Estimation PR20 Torch

---

3DFace

1. THF3D: Towards High-Fidelity 3D Face Reconstruction from In-the-Wild Images Using Graph Convolutional Networks PR20 Graph
2. AvatarMe: Realistically Renderable 3D Facial Reconstruction "in-the-wild" Project

---

Matching

1. RPMNet: Robust Point Matching using Learned Features PR20 Code
2. Learning Multiview 3D Point Cloud Registration PR20
3. D3Feat: Joint Learning of Dense Detection and Description of 3D Local Features PR20 TF
4. End-to-End Learning Local Multi-view Descriptors for 3D Point Clouds PR20

---

Keypoints

1. PVN3D: A Deep Point-wise 3D Keypoints Voting Network for 6DoF Pose Estimation PR20 Torch Video
2. KeypointNet: A Large-scale 3D Keypoint Dataset Aggregated from Numerous Human Annotations PR20

---

Layout

1. Total3DUnderstanding: Joint Layout, Object Pose and Mesh Reconstruction for Indoor Scenes from a Single Image PR20
2. IntelligentHome3D: Automatic 3D-House Design from Linguistic Descriptions Only PR20

---

Depth

1. Novel View Synthesis of Dynamic Scenes with Globally Coherent Depths from a Monocular Camera PR20
2. Towards Better Generalization: Joint Depth-Pose Learning without PoseNet PR20 Torch [PoseEstimation]
3. Depth Sensing Beyond LiDAR Range PR20

---

Surface

1. Articulation-aware Canonical Surface Mapping PR20 Torch Project
2. Deep 3D Capture: Geometry and Reflectance from Sparse Multi-View Images PR20
3. Where Does It End? -- Reasoning About Hidden Surfaces by Object Intersection Constraints PR20

---

Texture

1. Learning to Transfer Texture from Clothing Images to 3D Humans PR20 Code
2. TheVirtualTailor: Predicting Clothing in 3D as a Function of Human Pose, Shape and Garment Style PR20 Code
3. Learning to Dress 3D People in Generative Clothing PR20
4. Towards Photo-Realistic Virtual Try-On by Adaptively Generating $\leftrightarrow $ Preserving Image Content PR20

---

Graph

1. OccuSeg: Occupancy-aware 3D Instance Segmentation PR20 C/S
2. PointGNN: Graph Neural Network for 3D Object Detection in a Point Cloud PR20 TF Detection
3. HOPENet: A Graph-based Model for Hand-Object Pose Estimation PR20 Project Torch Account Pose Estimation
4. THF3D: Towards High-Fidelity 3D Face Reconstruction from In-the-Wild Images Using Graph Convolutional Networks PR20 Reconstruction
5. GridGCN: Grid-GCN for Fast and Scalable Point Cloud Learning PR20 Account C/S

---

Datasets

1. SPARE3D: A Dataset for Spatial Reasoning on Three-View Line Drawings PR20 Project
2. Cross-View Tracking for Multi-Human 3D Pose Estimation at over 100 FPS PR20
3. IntrA: 3D Intracranial Aneurysm Dataset for Deep Learning PR20

---

Others

1. StyleRig: Rigging StyleGAN for 3D Control over Portrait Images PR20 Project
2. Neural Pose Transfer By Spatially Adaptive Instance Normalization PR20 Code Account
3. GeoDA: A Geometric Framework for Black-box Adversarial Attacks PR20
4. Neural Contours: Learning to Draw Lines from 3D Shapes PR20
5. Robust 3D Self-portraits in Seconds PR20 Oral
6. Leveraging 2D Data to Learn Textured 3D Mesh Generation PR20 Oral
7. X3D: Expanding Architectures for Efficient Video Recognition PR20 Oral Code
8. 3D Photography using Context-aware Layered Depth Inpainting PR20 Code

---