Machine Learning Meets Geometry

CSE291-I00 - Winter 2021


Schedule


Section 1

Theories of Geometry

1/5
Curve Theory overview of the course, logistics, curve theory
1/7
Surface Theory differential map, normal curvature, principal curvature
1/12
Surface Theory (II) shape operator, first fundamental form, isometry, fundamental theorem of surfaces
1/14
Mesh and Point Cloud polygonal mesh, point cloud
1/19
Rotation and SO(3) rotation matrix, euler angle, angle-axis, quaternion

Section 2

3D Deep Learning

Section 2.1

3D Reconstruction

1/21
Learning-based MVS learning-based MVS, NeRF
1/26
Single Image to 3D EMD, Chamfer, mesh deformation

Section 2.2

3D Data Understanding

1/28
3D Backbone Networks Volumetric CNN, PointNet
2/2
3D Detection guest lecture by Dr. Charles Qi. Frustum PointNet, PointPillar, VoteNet
2/4
3D Instance Segmentation top-down approach, bottom-up approach
2/9
6D Pose Estimation ICP, Umeyama's method, direct method
2/11
6D Pose Estimation (II) indirect approach, DenseFusion, PVN3D, NOCS
2/18
Intrinsics-based Analysis geodesic distance, dijkstra's algorithm for geodesics, learning-based method for geodesics, applications

Section 2.3

Structured 3D Learning

2/16
Part-based 3D Analysis guest lecture by Kaichun Mo
2/23
Zero-shot 3D Understanding correspondence-based part discovery, learning to group
2/25
Deformation Models surface deformation, space deformation, skeleton skinning
3/2
3D Human Body and Behaviour guest lecture by Prof. Angjoo Kanazawa (password protected)

Section 2.4

Geometry Processing and Collection Analysis

3/4
Surface Reconstruction explicit and implicit method for surface reconstruction
3/9
Correspondences and Cycle Consistency guest lecture by Prof. Qixing Huang
3/11
Mesh Processing misc mesh processing problems and a brief intro to the course next quarter