Predicting Occupation from Images by Combining Face and Body Context Information
碩士 === 國立中正大學 === 資訊工程研究所 === 103 === Facial images embed age, gender, and other rich information that is implicitly related to occupation. In this work, we advocate that occupation prediction from a single facial image is a doable computer vision problem. We first extract visual features from multi...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | en_US |
Published: |
2015
|
Online Access: | http://ndltd.ncl.edu.tw/handle/92504697562122918216 |
id |
ndltd-TW-102CCU00392113 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-102CCU003921132016-07-31T04:21:41Z http://ndltd.ncl.edu.tw/handle/92504697562122918216 Predicting Occupation from Images by Combining Face and Body Context Information 結合人臉及身體脈絡資訊於影像中預測職業 Chih-Hao Chiu 邱志豪 碩士 國立中正大學 資訊工程研究所 103 Facial images embed age, gender, and other rich information that is implicitly related to occupation. In this work, we advocate that occupation prediction from a single facial image is a doable computer vision problem. We first extract visual features from multiple levels of patches and construct occupation descriptors by locality-constrained linear coding. To avoid the curse of dimensionality and overfitting, a boost strategy called multi-feature SVM is used to integrate features. Intra-class and inter-class visual variations are jointly considered in the boosting framework to further improve performance. We also employ the same framework to body context, and integrate heterogeneous information to predict occupation. In the evaluation, we verify effectiveness of predicting occupation from face, demonstrate promising performance obtained by combining face and body information, and investigate the influence of model parameters on prediction performance. Wei-Ta Chu 朱威達 2015 學位論文 ; thesis 50 en_US |
collection |
NDLTD |
language |
en_US |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立中正大學 === 資訊工程研究所 === 103 === Facial images embed age, gender, and other rich information that is implicitly related to occupation. In this work, we advocate that occupation prediction from a single facial image is a doable computer vision problem. We first extract visual features from multiple levels of patches and construct occupation descriptors by locality-constrained linear coding. To avoid the curse of dimensionality and overfitting, a boost strategy called multi-feature SVM is used to integrate features. Intra-class and inter-class visual variations are jointly considered in the boosting framework to further improve performance. We also employ the same framework to body context, and integrate heterogeneous information to predict occupation. In the evaluation, we verify effectiveness of predicting occupation from face, demonstrate promising performance obtained by combining face and body information, and investigate the influence of model parameters on prediction performance.
|
author2 |
Wei-Ta Chu |
author_facet |
Wei-Ta Chu Chih-Hao Chiu 邱志豪 |
author |
Chih-Hao Chiu 邱志豪 |
spellingShingle |
Chih-Hao Chiu 邱志豪 Predicting Occupation from Images by Combining Face and Body Context Information |
author_sort |
Chih-Hao Chiu |
title |
Predicting Occupation from Images by Combining Face and Body Context Information |
title_short |
Predicting Occupation from Images by Combining Face and Body Context Information |
title_full |
Predicting Occupation from Images by Combining Face and Body Context Information |
title_fullStr |
Predicting Occupation from Images by Combining Face and Body Context Information |
title_full_unstemmed |
Predicting Occupation from Images by Combining Face and Body Context Information |
title_sort |
predicting occupation from images by combining face and body context information |
publishDate |
2015 |
url |
http://ndltd.ncl.edu.tw/handle/92504697562122918216 |
work_keys_str_mv |
AT chihhaochiu predictingoccupationfromimagesbycombiningfaceandbodycontextinformation AT qiūzhìháo predictingoccupationfromimagesbycombiningfaceandbodycontextinformation AT chihhaochiu jiéhérénliǎnjíshēntǐmàiluòzīxùnyúyǐngxiàngzhōngyùcèzhíyè AT qiūzhìháo jiéhérénliǎnjíshēntǐmàiluòzīxùnyúyǐngxiàngzhōngyùcèzhíyè |
_version_ |
1718366716673654784 |