Bilayer Part-based Model for Facial Landmark Detection and Pose Estimation
碩士 === 國立臺灣科技大學 === 機械工程系 === 103 === Tree Structured model (TSM) is proven effective for face detection, landmark localization and pose estimation. It is a rare approach that can solve all three issues using one single unified model. However, it can be too slow to handle real-time applications beca...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
2015
|
Online Access: | http://ndltd.ncl.edu.tw/handle/14157924061690028082 |
id |
ndltd-TW-103NTUS5489014 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-103NTUS54890142016-11-06T04:19:26Z http://ndltd.ncl.edu.tw/handle/14157924061690028082 Bilayer Part-based Model for Facial Landmark Detection and Pose Estimation 基於雙層式元件模型之人臉地標點定位與角度估測 Kai-Hsiang Chang 張凱翔 碩士 國立臺灣科技大學 機械工程系 103 Tree Structured model (TSM) is proven effective for face detection, landmark localization and pose estimation. It is a rare approach that can solve all three issues using one single unified model. However, it can be too slow to handle real-time applications because of the heavy computation involved. Besides, it cannot detect faces less than 80x80 in size. A bilayer structure, coined Bilayer Tree Structure Model(BTSM), is proposed in this study to solve these two issues. The BTSM has a downscaled model with fewer parts and trained on down-scaled samples, and therefore, can detect faces as small as 50x50. When the down-scaled model finds faces of sufficient sizes, it would activate a full-scaled model to locate more landmarks without performing convolution through the image pyramid. Compared on various databases, the BTSM can be 30x faster than the original TSM, while keeping almost all advantages of TSM the same. Gee-Sern Hsu 徐繼聖 2015 學位論文 ; thesis 76 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立臺灣科技大學 === 機械工程系 === 103 === Tree Structured model (TSM) is proven effective for face detection, landmark
localization and pose estimation. It is a rare approach that can solve all three issues using one single unified model. However, it can be too slow to handle real-time applications because of the heavy computation involved. Besides, it cannot detect faces less than 80x80 in size. A bilayer structure, coined Bilayer Tree Structure Model(BTSM), is proposed in this study to solve these two issues. The BTSM has a downscaled model with fewer parts and trained on down-scaled samples, and therefore, can detect faces as small as 50x50. When the down-scaled model finds faces of sufficient sizes, it would activate a full-scaled model to locate more landmarks without performing convolution through the image pyramid. Compared on various databases, the BTSM can be 30x faster than the original TSM, while keeping almost all advantages of TSM the same.
|
author2 |
Gee-Sern Hsu |
author_facet |
Gee-Sern Hsu Kai-Hsiang Chang 張凱翔 |
author |
Kai-Hsiang Chang 張凱翔 |
spellingShingle |
Kai-Hsiang Chang 張凱翔 Bilayer Part-based Model for Facial Landmark Detection and Pose Estimation |
author_sort |
Kai-Hsiang Chang |
title |
Bilayer Part-based Model for Facial Landmark Detection and Pose Estimation |
title_short |
Bilayer Part-based Model for Facial Landmark Detection and Pose Estimation |
title_full |
Bilayer Part-based Model for Facial Landmark Detection and Pose Estimation |
title_fullStr |
Bilayer Part-based Model for Facial Landmark Detection and Pose Estimation |
title_full_unstemmed |
Bilayer Part-based Model for Facial Landmark Detection and Pose Estimation |
title_sort |
bilayer part-based model for facial landmark detection and pose estimation |
publishDate |
2015 |
url |
http://ndltd.ncl.edu.tw/handle/14157924061690028082 |
work_keys_str_mv |
AT kaihsiangchang bilayerpartbasedmodelforfaciallandmarkdetectionandposeestimation AT zhāngkǎixiáng bilayerpartbasedmodelforfaciallandmarkdetectionandposeestimation AT kaihsiangchang jīyúshuāngcéngshìyuánjiànmóxíngzhīrénliǎndebiāodiǎndìngwèiyǔjiǎodùgūcè AT zhāngkǎixiáng jīyúshuāngcéngshìyuánjiànmóxíngzhīrénliǎndebiāodiǎndìngwèiyǔjiǎodùgūcè |
_version_ |
1718391666157551616 |