/
Structured Forests for Structured Forests for

Structured Forests for - PowerPoint Presentation

mitsue-stanley
mitsue-stanley . @mitsue-stanley
Follow
591 views
Uploaded On 2015-09-20

Structured Forests for - PPT Presentation

Fast Edge Detection Piotr Dollár and Larry Zitnick what defines an edge Brightness Color Texture Parallelism Continuity Symmetry Let the data speak 1 Accuracy 2 Speed I data driven edge detection ID: 134917

edge structured forests detection structured edge detection forests ods split entropy doll

Share:

Link:

Embed:

Download Presentation from below link

Download Presentation The PPT/PDF document "Structured Forests for" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.


Presentation Transcript

Slide1

Structured Forests for

Fast Edge Detection

Piotr Dollár and Larry ZitnickSlide2

what defines an edge?

Brightness

ColorTextureParallelismContinuitySymmetry …

Let the data speak.Slide3

1. Accuracy

2. SpeedSlide4

I. data driven edge detectionSlide5

edge detection as classification

Supervised Learning of Edges and Object Boundaries

CVPR 2006, Piotr Dollár, Zhuowen Tu, Serge Belongie

positives

Hard!

{ 0, 1 }Slide6

edge have structureSlide7

sketch tokens

Sketch Tokens, CVPR 2013. Joseph Lim

, C. Zitnick, and P. DollárSlide8

random forestsSlide9

upgrading the output space

{ 0, 1 }{

… }

dimensionality

2

2

256

151Slide10

II. structured edge learningSlide11

structured forests

Structured

Class-Labels in Random Forests for Semantic Image Labelling,

ICCV 2011,

P. Kontschieder

,

S. Rota

Bulò

, H.

Bischof

, M.

PelilloSlide12

tree trainingSlide13

node training

high entropy split

low entropy split

Slide14

how to train?

bad split

good split

?

?Slide15

?

?

cluster

minimize entropySlide16

III. structured edge detectionSlide17

structured forestsSlide18

sliding window detectorSlide19

sliding window detector

pixel output

structured output Slide20

multiscale detectionSlide21

½ x

2 x

1 x

multiscale detection

+

+Slide22

IV. resultsSlide23

SS=single-scale

MS=multi-scale T=# trees

6Hz

SS T=1

SS T=4

MS T=4

ODS = 0.72

ODS = 0.74

ODS = 0.73

60Hz

30Hz

gPb ODS=.73

gPb ODS=.73

FPS ≈ 1/240 HzSlide24
Slide25
Slide26
Slide27
Slide28

thanks!

source code available online