Volumetric and Multi-View CNNs for Object Classification on 3D Data - - PowerPoint PPT Presentation

volumetric and multi view cnns for object classification
SMART_READER_LITE
LIVE PREVIEW

Volumetric and Multi-View CNNs for Object Classification on 3D Data - - PowerPoint PPT Presentation

Volumetric and Multi-View CNNs for Object Classification on 3D Data Charles R. Qi*, Hao Su*, Matthias Niener, Angela Dai, MengyuanYan, Leonidas J.Guibas Rich Applications of 3D Augmented Robot Reality Perception 3D Representations for


slide-1
SLIDE 1

Volumetric and Multi-View CNNs for Object Classification on 3D Data

Charles R. Qi*, Hao Su*, Matthias Nießner, Angela Dai, MengyuanYan, Leonidas J.Guibas

slide-2
SLIDE 2

Rich Applications of 3D

Augmented Reality Robot Perception

slide-3
SLIDE 3

3D Representations for Generic Object Classification

3DShapeNets by Z. Wu et

  • al. CVPR 15

VoxNet by D. Maturana et

  • al. IEEE/RSJ 15

MVCNN by H. Su et al. ICCV 15 DeepPano by B. Shi et al. IEEE/SPL 15

Volumetric Multi-Views

slide-4
SLIDE 4

Volumetric CNNs Revisited

3DShapeNets by Z. Wu et

  • al. CVPR 15

Volumetric CNNs

slide-5
SLIDE 5

Multi-View CNNs Revisited

MVCNN by H. Su et al. ICCV 15

Multi-View CNNs

slide-6
SLIDE 6

Shape Classification Results Revisited

3DShapeNets Wu et al. MVCNN Su et al. 70 75 80 85 90 95

77.3% 90.1%

slide-7
SLIDE 7

Shape Classification Results Revisited

3DShapeNets Wu et al. MVCNN Su et al. 70 75 80 85 90 95

77.3% 90.1%

Big gap between volumetric and multi-view based methods

Why?

slide-8
SLIDE 8

Cause 1: Architecture and Engineering

LeNet, 1998 AlexNet, 2012

slide-9
SLIDE 9

Cause 1: Architecture and Engineering

LeNet, 1998 AlexNet, 2012 3DShapeNets, 2015

slide-10
SLIDE 10

Cause 2: Resolution

224x224 Images Multi-View CNNs

MVCNN Su et al.

slide-11
SLIDE 11

Cause 2: Resolution

224x224 Images 30x30x30 Volumes Volumetric CNNs

3DShapeNets Wu et al.

Multi-View CNNs

MVCNN Su et al.

slide-12
SLIDE 12

Diagnosis of Causes: Variable Control

  • Same resolution, study architectures
  • Same architecture, look into resolutions
slide-13
SLIDE 13

Sphere Rendering

Occupancy Grid 30x30x30 Polygon Mesh Image 224x224

slide-14
SLIDE 14

Sphere Rendering

Occupancy Grid 30x30x30 Polygon Mesh Image 224x224 Same “3D Resolution”

slide-15
SLIDE 15

Investigation into Architecture

3D CNN Multi-View Image CNN Different Architecture

Sphere Rendering Images Occupancy Grid Volumes

Same 3D Resolution (30x30x30)

slide-16
SLIDE 16

CNNs with Same 3D Resolution Inputs

72 74 76 78 80 82 84 86 88

MVCNN with Sphere Rendering Images 3DShapeNets Wu et al. Shape Classification Accuracy

slide-17
SLIDE 17

Novel 3D CNN Architectures

 3D NIN with Subvolume Supervision Push Harder for Learning Better!

slide-18
SLIDE 18

Novel 3D CNN Architectures

Anisotropic Probing Network

slide-19
SLIDE 19

Results of Our Novel 3D CNNs

72 74 76 78 80 82 84 86 88

MVCNN with Sphere Rendering Images 3DShapeNets Wu et al. Ours 3D CNN Shape Classification Accuracy

slide-20
SLIDE 20

72 74 76 78 80 82 84 86 88

MVCNN with Sphere Rendering Images 3DShapeNets Wu et al. Ours 3D CNN

Results of Our Novel 3D CNNs

Closed the Gap under same 3D Resolution

Shape Classification Accuracy

slide-21
SLIDE 21

Investigation into Resolution

Multi-View Image CNN 3D CNN Multi-View Image CNN

Standard Rendering Images Sphere Rendering Images 30x30x30 Volume

Same Architecture Different 3D Resolution

slide-22
SLIDE 22

Performance Trend wrt 3D Resolution

82 84 86 88 90 92 94 50 100 150 200 250

Accuracy (%) 3D Resolution

MVCNN-Sphere

slide-23
SLIDE 23

Performance Trend wrt 3D Resolution

82 84 86 88 90 92 94 50 100 150 200 250

Accuracy (%) 3D Resolution

MVCNN-Sphere Our 3D CNN

slide-24
SLIDE 24

Generalization to Real Scans

Real Scan Dataset 243 objects 12 categories

Shape retrieval on scan data

slide-25
SLIDE 25

Code and Data Available Online! Welcome to Our Poster #38!

Volumetric and Multi-View CNNs for Object Classification on 3D Data

http://graphics.stanford.edu/projects/3dcnn/