• No results found

Cover Page The handle http://hdl.handle.net/1887/45135 holds various files of this Leiden University dissertation. Author: Wu, S. Title: Large scale visual search Issue Date: 2016-12-22

N/A
N/A
Info
Downloaden
Protected

Academic year: 2021

Share "Cover Page The handle http://hdl.handle.net/1887/45135 holds various files of this Leiden University dissertation. Author: Wu, S. Title: Large scale visual search Issue Date: 2016-12-22"

Copied!
2
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Download het nu ( 2 pagina )

Hele tekst

(1)

Cover Page

The handle http://hdl.handle.net/1887/45135 holds various files of this Leiden University dissertation.

Author: Wu, S.

Title: Large scale visual search

Issue Date: 2016-12-22

(2)

Large Scale Visual Search

Song Wu

Is “deep learning” the future of, well, everything?

conv1

conv1_1

conv2

conv2_1

conv3

conv3_1

Integration layer

Fully connected layer pool1

weight matrix 1

pool2

weight matrix 2

pool3

weight matrix 3

weight matrix 4

Softmax (a classifier)

Lar ge Scale V isual Search Song W u

INVITATION

You are kindly invited to attend the defence of my PhD

dissertation:

Large Scale Visual Search

on Thursday 22 December, 2016

At 16:15 PM

In the Academiegebouw Rapenburg 73, Leiden

Afterwards, a reception will be held in Woo Ping restaurant Diefsteeg 15, 2311 TS Leiden

Song Wu

s.wu@liacs.leidenuniv.nl Paranymphs:

Yanming Guo Zhan Xiong

Deep binary code

…..

.

…... Bit-scalable

A pre-trained CNN Query image

Image search results

Late fusion Deep binary code

…...

…... Bit-scalable Deep binary code

…...

…... Bit-scalable

……

Database

Final results

……

……

……

……

Large Scale Visual Search

Song Wu

Is “deep learning” the future of, well, everything?

conv1

conv1_1

conv2

conv2_1

conv3

conv3_1

Integration layer Fully connected layer pool1

weight matrix 1

pool2

weight matrix 2

pool3

weight matrix 3 weight matrix 4

Softmax (a classifier)

Lar ge Scale V isual Search Song W u

11 1 11

1 1

1 1 1

1 1 1

1 1 1 1 11 1 1

0 00

0 0

0 0 0 0 0

0

0

0 0 0

INVITATION

You are kindly invited to attend the defence of my PhD

dissertation:

Large Scale Visual Search

on Thursday 22 December, 2016

At 16:15 PM In the Academiegebouw

Rapenburg 73, Leiden

Afterwards, a reception will be held in Woo Ping restaurant Diefsteeg 15, 2311 TS Leiden

Song Wu s.wu@liacs.leidenuniv.nl

Paranmphs: Yanming Guo

Zhan Xiong

Deep binary code

…..

…..

Bit-scalable

A pre-trained CNN Query image

Image search results

Late fusion Deep binary code

…..

Bit-scalable …..

Deep binary code

…..

Bit-scalable …..

……

Database

Final results

……

……

……

……

11 1 11

1 1

1 1 1

1 1 1

1 1 1 1 11 1 1

0 00

0 0

0 0 0 0 0

0

0

0 0 0

Deep binary code

…..

….. Bit-scalable

A pre-trained CNN Query image

Image search results

Late fusion Deep binary code

….. .

….. . Bit-scalable Deep binary code

….. .

….. . Bit-scalable

…… Database

Final results

……

……

……

……

9 789463 321174

Deep binary code

…...

…..

. Bit-scalable

A pre-trained CNN Query image

Image search results

Late fusion Deep binary code

…..

.

…..

. Bit-scalable

Deep binary code

…...

…... Bit-scalable

……

Database

Final results

……

……

……

……

Large Scale Visual Search

Song Wu

Is “deep learning” the future of, well, everything?

conv1

conv1_1

conv2

conv2_1

conv3

conv3_1

Integration layer

Fully connected layer pool1

weight matrix 1

pool2

weight matrix 2

pool3

weight matrix 3

weight matrix 4

Softmax (a classifier)

Lar ge Scale V isual Search Song W u

1 1 1 1 1

1 1

1 1 1

1 1 1

1 1 1 1 1 1 1 1

0 0 0

0 0

0 0 0 0 0

0

0

0 0 0

INVITATION

You are kindly invited to attend the defence of my PhD

dissertation:

Large Scale Visual Search

on Thursday 22 December, 2016

At 16:15 PM

In the Academiegebouw Rapenburg 73, Leiden

Afterwards, a reception will be held in Woo Ping restaurant Diefsteeg 15, 2311 TS Leiden

Song Wu

s.wu@liacs.leidenuniv.nl Paranymphs:

Yanming Guo Zhan Xiong

Deep binary code

…..

Bit-scalable …..

A pre-trained CNN Query image

Image search results

Late fusion Deep binary code

…..

Bit-scalable …..

Deep binary code

…..

Bit-scalable …..

……

Database

Final results

……

……

……

……

1 1 1 1 1

1 1

1 1 1

1 1 1

1 1 1 1 1 1 1 1

0 0 0

0 0

0 0 0 0 0

0

0

0 0 0

Deep binary code

…..

.

…... Bit-scalable

A pre-trained CNN Query image

Image search results

Late fusion Deep binary code

…...

…... Bit-scalable Deep binary code

…...

…... Bit-scalable

……

Database

Final results

……

……

……

……

Large Scale Visual Search

Song Wu

Is “deep learning” the future of, well, everything?

conv1

conv1_1

conv2

conv2_1

conv3

conv3_1

Integration layer Fully connected layer pool1

weight matrix 1

pool2

weight matrix 2

pool3

weight matrix 3 weight matrix 4

Softmax (a classifier)

Lar ge Scale V isual Search Song W u

11 1 11

1 1

1 1 1

1 1 1

1 1 1 1 11 1 1

0 00

0 0

0 0 0 0 0

0

0

0 0 0

INVITATION

You are kindly invited to attend the defence of my PhD

dissertation:

Large Scale Visual Search

on Thursday 22 December, 2016

At 16:15 PM In the Academiegebouw

Rapenburg 73, Leiden

Afterwards, a reception will be held in Woo Ping restaurant Diefsteeg 15, 2311 TS Leiden

Song Wu s.wu@liacs.leidenuniv.nl

Paranmphs: Yanming Guo

Zhan Xiong

Deep binary code

…..

…..

Bit-scalable

A pre-trained CNN Query image

Image search results

Late fusion Deep binary code

…..

Bit-scalable …..

Deep binary code

…..

Bit-scalable …..

……

Database

Final results

……

……

……

……

11 1 11

1 1

1 1 1

1 1 1

1 1 1 1 11 1 1

0 00

0 0

0 0 0 0 0

0

0

0 0 0

Deep binary code

…..

….. Bit-scalable

A pre-trained CNN Query image

Image search results

Late fusion Deep binary code

….. .

….. . Bit-scalable Deep binary code

….. .

….. . Bit-scalable

…… Database

Final results

……

……

……

……

9 789463 321174

Referenties

Download het nu ( PDF - 2 pagina - 2.08 MB )

GERELATEERDE DOCUMENTEN

Cover Page The handle http://hdl.handle.net/1887/45135 holds various files of this Leiden University dissertation. Author: Wu, S. Title: Large scale visual search Issue Date: 2016-12-22

Convolutional neural networks based deep hash learning: In order to achieve efficient large scale image search, the high performance of the supervised deep hashing model appears to

Cover Page The handle http://hdl.handle.net/1887/45135 holds various files of this Leiden University dissertation. Author: Wu, S. Title: Large scale visual search Issue Date: 2016-12-22

In: Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern

Cover Page The handle http://hdl.handle.net/1887/45135 holds various files of this Leiden University dissertation. Author: Wu, S. Title: Large scale visual search Issue Date: 2016-12-22

The evaluation results showed that the binary string descriptor based BoW model has low memory requirements for vocabulary storage and competi- tive performance compared with

Cover Page The handle http://hdl.handle.net/1887/45135 holds various files of this Leiden University dissertation. Author: Wu, S. Title: Large scale visual search Issue Date: 2016-12-22

De evaluatie-resultaten laten zien dat deze aanpak weinig geheugen gebruikt voor vocabulaire opslag en com- petetief presteert vergeleken met BoW afbeeldingsrepresentaties gebaseerd

Cover Page The handle http://hdl.handle.net/1887/45135 holds various files of this Leiden University dissertation. Author: Wu, S. Title: Large scale visual search Issue Date: 2016-12-22

Many thanks to my friends in the Netherlands: Zhan Xiong, Junlin He, Boyang Liu, Qinyin Hu, Shengfa Miao, Di liu, Fuyu Cai, Yuanhao Guo, Xiaoqing Tang, Enrique Larios Vargas,

Cover Page The handle http://hdl.handle.net/1887/20443 holds various files of this Leiden University dissertation. Author:

At this point, I should also thank colleagues and staff at the Turkish Studies Department of Leiden University, the International Institute of Social History, which

Cover Page The handle http://hdl.handle.net/1887/20443 holds various files of this Leiden University dissertation. Author:

The results derived from these textual and discursive practices will be identified in the hegemonic analysis of the two military intervention processes, with a special focus on the

Cover Page The handle http://hdl.handle.net/1887/20443 holds various files of this Leiden University dissertation. Author:

The static character, theoretical limitations, and geographical dependency of Mardin’s center- periphery analysis are overcome by Gramscian hegemony’s analytical and theoretical depth