View Planning for Object Recognition Gabriel Oliveira and Volkan - - PowerPoint PPT Presentation

▶

Jan 07, 2024 483 likes •796 views

View Planning for Object Recognition Gabriel Oliveira and Volkan Isler RSN Lab Motivation 2/30 Objective Cloud-Based (Active) Object recognition Goal: Find the minimum amount of views for recognition 3/30 Problem Definition 4/30

SLIDE 1

View Planning for Object Recognition

Gabriel Oliveira and Volkan Isler RSN Lab

SLIDE 2

Motivation

2/30

SLIDE 3

Objective

Cloud-Based (Active) Object

recognition

Goal: Find the minimum amount of

views for recognition

3/30

SLIDE 4

Problem Definition

4/30

SLIDE 5

System Overview

5/30

SLIDE 6

Recognition

Recognition module
[Vincze et al. 2012]:

l Segmentation (RANSAC) l Descriptor (ESF) l Matching (KNN) l Merging (Max all views)

6/30

SLIDE 7

Viewpoints

Open loop approach:

l No prior Knowledge about the next view

Approximation of Edge Based Best

Next View approach [Abidi et al. 2000]:

l Explore areas of occlusion l Approximate the three first views to be

pairwise orthogonal

7/30

SLIDE 8

Viewpoints

Empirical Upper Bound the

number of views:

l 4 views in a plane:

l All views are orthogonal to its 2 closest

neighbors 8/30

SLIDE 9

Experiments

Dataset
Time performance

l Communication l System Bottleneck (segmentation)

Recognition Results
Distribution of Viewpoints

9/30

SLIDE 10

Experiments

Dataset
Time performance

l Communication l System Bottleneck (segmentation)

Recognition Results
Distribution of Viewpoints

10/30

SLIDE 11

Experiments - Setup

11/30

SLIDE 12

Experiments

Used Dataset

12/30

SLIDE 13

Experiments

Dataset
Time performance

l Communication l System Bottleneck (segmentation)

Recognition Results
Distribution of Viewpoints

13/30

SLIDE 14

Experiments

Communication Results

Method Mean Standard Deviation Size of cloud Sent Transmission without Passthrough filter 10.35 fps 2.28 105 Kb from 4500 Kb original size Transmission with Passthrough filter from 1.0 to 3.5 meters 6.55 fps 1.40 87 Kb from 4500 Kb original size

14/30

SLIDE 15

Experiments

Dataset
Time performance

l Communication l System Bottleneck (segmentation)

Recognition Results
Distribution of Viewpoints

15/30

SLIDE 16

Experiments

Segmentation

Segmentation # of

bjects (frame-rate)

Minimum (ms) Maximum (ms) 1 object (~3.4 fps) 270 310 2 object (~2.6 fps) 355 400 3 object (~1.9 fps) 500 530

16/30

SLIDE 17

Experiments

Dataset
Time performance

l Communication l System Bottleneck (segmentation)

Recognition Results
Distribution of Viewpoints

17/30

SLIDE 18

Experiments Recognition

Recognition from 0, 90, 180 and 270

degrees

Fused recognition based on multiple

views

18/30

SLIDE 19

Experiments Recognition

90 180 270

SLIDE 20

Experiments Recognition

Highest Values

20/30

SLIDE 21

Experiments

Dataset
Time performance

l Communication l System Bottleneck (segmentation)

Recognition Results
Distribution of Viewpoints

21/30

SLIDE 22

Distribution of Viewpoints

Representative views of classes that

present significant fluctuations:

l Stapler, Cap, Keyboard and Car

22/30

SLIDE 23

Viewpoints Distribution

Stapler

23/30

SLIDE 24

Viewpoints Distribution

24/30

SLIDE 25

Viewpoints Distribution

Keyboard

25/30

SLIDE 26

Viewpoints Distribution

26/30

SLIDE 27

Conclusions and Future Works

Four views show promising

results

Our goal is to prove this

analytically

System present high recognition

rates to most of the objects

27/30

SLIDE 28

Conclusions and Future Works

Test with larger datasets
Refine or propose new approaches

to:

l Segmentation l Partial Viewpoint generation for training

28/30

SLIDE 29

Thanks

Contact:

 olvieira@cs.umn.edu l rsn.cs.umn.edu