Schedule

Schedule	Broad Area	Reading List
Jan 12	Course Intro + intro to 3D representations
Jan 19	1.Representations of 3D data in Modern ML	PointNet and PointNet++		Slides (Dylan Turpin)
		Local Deep Implicit Functions for 3D Shape		Slides (Jun Gao)
		Optional: Points2Surf, NOCS, DeepSDF
Jan 26	2.Learning: 3D Shape Modelling	PointConv
		Dynamic Graph CNN for Learning on Point Clouds	Presentation (Mustafa Haiderbhai)	Slides
		KPConv	Presentation (Bin Yang)	Slides
		DeepSDF	Presentation (Tianchang Shen)	Slides
		Optional: PCT: Point Cloud Transformer, Learning Deformable Tetrahedral Meshes for 3D Reconstruction, PIE-NET
Feb 2	3.Computational Efficiency in Processing of 3D Data + Scene Understanding	Guest speaker: Dr. Chris Choy (Nvidia) presenting MinowskiNets
		Virtual Multi-view Fusion for 3D Semantic Segmentation	Presentation (Xiang Cao)	Slides
		3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans	Presentation (Haoping Xu)	Slides
		Optional: Unsupervised part representation by Flow Capsules, OccuSeg, SparseConvNet
Feb 9	4.Generative Modelling in 3D	Learning Generative Models of 3D Structures	Presentation (Tao Wu)	Slides
		PolyGen	Presentation (Alexander Tessier)	Slides
		Learning Gradient Fields for Shape Generation	Presentation (Andrej Janda)
		ShapeAssembly	Presentation (Zhoujie Zhao)	Slides
		AtlasNet	Presentation (Varun Pandya)	Slides
		Optional: StructureNet
Feb 16	5.Differentiable Rendering	Differentiable Rendering: A Survey	Presentation (Ze Yang)	Slides
		Differentiable Monte Carlo Ray Tracing through Edge Sampling	Presentation (Jingkang Wang)
		Multiview Neural Surface Reconstruction with Implicit Lighting and Material	Presentation (Wenzhi Guo)	Slides
		Differentiable Volumetric Rendering	Presentation (Sara Sabour)	Slides
		Optional: Soft Rasterizer: A Differentiable Renderer for Image-based 3D Reasoning, Learning to Predict 3D Objects with an Interpolation-based Differentiable Renderer
Feb 23	6.Neural Rendering NeRF	Neural Volume Rendering: NeRF And Beyond
		NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis
		NeRF in the Wild: Neural Radiance Fields for Unconstrained Photo Collections	Presentation (Gary Leung)	Slides
		Scene Representation Networks: Continuous 3D-Structure-Aware Neural Scene Representations	Presentation (Shayan Shekarforoush)	Slides
		Neural Reflectance Fields for Appearance Acquisition	Presentation (Zian Wang)	Slides
		Optional: NeRF Explosion 2020
Mar 2	7.NeRF Applications	NSVF: Neural Sparse Voxel Fields	Presentation (Tianxing Li)	Slides
		NeRFies: Deformable Neural Radiance Fields	Presentation (Yun-Chun Chen)	Slides
		iNeRF: Inverting Neural Radiance Fields for Pose Estimation	Presentation (Bin Shi)	Slides
		Optional: NASA: Neural Articulated Shape Approximation, GRF: Learning a General Radiance Field for 3D Scene Representation and Rendering
Mar 9	8.Equivariance and Invariance	Isometric Transformation Invariant and Equivariant Graph Convolutional Networks
		Gauge Equivariant Mesh CNNs	Presentation (Otman Benchekroun)	Slides
		On Learning Sets of Symmetric Elements	Presentation (Dmitrii Shubin)	Slides
		CNNs on Surfaces using Rotation-Equivariant Features	Presentation (Shichen Lu)	Slides
		Neural Mesh Flow: 3D Manifold Mesh Generation via Diffeomorphic Flows
Mar 16	9.Unsupervised 3D learning	Unsupervised Learning of Probably Symmetric Deformable 3D Objects from Images in the Wild	Presentation (Brendan Kolisnik)	Slides
		Canonical Capsules: Unsupervised Capsules in Canonical Pose	Presentation (Ioannis Xarchakos	Slides
		PointContrast: Unsupervised Pre-training for 3D Point Cloud Understanding
		Learning Delaunay Surface Elements for Mesh Reconstruction	Presentation (Brendan Duke)	Slides
		Optional: KeypointNet: Discovery of Latent 3D Keypoints via End-to-end Geometric Reasoning, Unsupervised Geometry-Aware Representation Learning for 3D Human Pose Estimation, Unsupervised part representation by Flow Capsules, Weakly-Supervised 3D Human Pose Learning via Multi-View Images in the Wild
Mar 23	10.Geometric Deep Learning Beyond Computer Vision	Guest speaker: Prof. Jonathan Kelly presenting Self-Supervised Deep Pose Corrections for Robust Visual Odometry
		Isometric Transformation Invariant and Equivariant Graph Convolutional Networks	Presentation (Sejin Kim)	Slides
		Implicit 3D Orientation Learning for 6D Object Detection from RGB Images
		NeMo: Neural Mesh Models of Contrastive Features for Robust 3D Pose Estimation	Presentation (William Ngo)	Slides
		Optional: 6-DOF GraspNet: Variational Grasp Generation for Object Manipulation, DenseFusion: 6D Object Pose Estimation by Iterative Dense Fusion, Dense Object Nets, KeyPoint Affordances for Category-Level Robotic Manipulation
Mar 30	11.3D vision in Robotics	Fast end-to-end learning on protein surfaces	Presentation (Youheng Ge)	Slides
		Relational inductive biases, deep learning, and graph networks	Presentation (Seung Wook Kim)	Slides
		LanczosNet: Multi-Scale Deep Graph Convolutional Networks	Presentation (Juan Carrillo)	Slides
		SIREN: Implicit Neural Representations with Periodic Activation Functions	Presentation (Zikun Chen)	Slides
		Optional: RigNet: Neural Rigging for Articulated Characters, 3DGV Seminar: Michael Bronstein -- Geometric Deep Learning
Apr 6	Project Presentations.
Apr 10	Take Home midterm.
Apr 12	Project Presentation Buffer.