Face detection through compact classifier using adaptive look-up-table

Yuya Hanai, Tadahiro Kuroda

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Face detection has been well studied in terms of accuracy and speed. However, required memory size reduction is still poorly studied, which is becoming a critical issue as platforms for face detection go tiny. In this paper, we propose a novel compact weak classifier using Adaptive Look-Up-Table (ALUT) for face detection on resource-constrained devices such as wearable sensor nodes. ALUT gives good approximation of log-likelihood [3] with fewer data, thus enabling the drastic reduction of classifier data size, keeping high accuracy and low computation cost. To generate an optimal ALUT, a new cost function called Weighted Sum of Absolute Difference (WSAD) is also proposed for further improvement. In our experiment, the classifier data size is reduced by 43% and the computation cost is reduced by 15% with same accuracy, compared to a conventional fixed LUT classifier.

Original languageEnglish
Title of host publicationProceedings - International Conference on Image Processing, ICIP
PublisherIEEE Computer Society
Pages1225-1228
Number of pages4
ISBN (Print)9781424456543
DOIs
Publication statusPublished - 2009
Event2009 IEEE International Conference on Image Processing, ICIP 2009 - Cairo, Egypt
Duration: 2009 Nov 72009 Nov 10

Other

Other2009 IEEE International Conference on Image Processing, ICIP 2009
CountryEgypt
CityCairo
Period09/11/709/11/10

Fingerprint

Face recognition
Classifiers
Sensor nodes
Cost functions
Costs
Data storage equipment
Experiments

Keywords

  • Dynamic programming
  • Object detection
  • Signal partitioning
  • Wearable computing

ASJC Scopus subject areas

  • Software
  • Computer Vision and Pattern Recognition
  • Signal Processing

Cite this

Hanai, Y., & Kuroda, T. (2009). Face detection through compact classifier using adaptive look-up-table. In Proceedings - International Conference on Image Processing, ICIP (pp. 1225-1228). [5413645] IEEE Computer Society. https://doi.org/10.1109/ICIP.2009.5413645

Face detection through compact classifier using adaptive look-up-table. / Hanai, Yuya; Kuroda, Tadahiro.

Proceedings - International Conference on Image Processing, ICIP. IEEE Computer Society, 2009. p. 1225-1228 5413645.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Hanai, Y & Kuroda, T 2009, Face detection through compact classifier using adaptive look-up-table. in Proceedings - International Conference on Image Processing, ICIP., 5413645, IEEE Computer Society, pp. 1225-1228, 2009 IEEE International Conference on Image Processing, ICIP 2009, Cairo, Egypt, 09/11/7. https://doi.org/10.1109/ICIP.2009.5413645
Hanai Y, Kuroda T. Face detection through compact classifier using adaptive look-up-table. In Proceedings - International Conference on Image Processing, ICIP. IEEE Computer Society. 2009. p. 1225-1228. 5413645 https://doi.org/10.1109/ICIP.2009.5413645
Hanai, Yuya ; Kuroda, Tadahiro. / Face detection through compact classifier using adaptive look-up-table. Proceedings - International Conference on Image Processing, ICIP. IEEE Computer Society, 2009. pp. 1225-1228
@inproceedings{aef303d1cc4c45e89ba0ec745cd410c7,
title = "Face detection through compact classifier using adaptive look-up-table",
abstract = "Face detection has been well studied in terms of accuracy and speed. However, required memory size reduction is still poorly studied, which is becoming a critical issue as platforms for face detection go tiny. In this paper, we propose a novel compact weak classifier using Adaptive Look-Up-Table (ALUT) for face detection on resource-constrained devices such as wearable sensor nodes. ALUT gives good approximation of log-likelihood [3] with fewer data, thus enabling the drastic reduction of classifier data size, keeping high accuracy and low computation cost. To generate an optimal ALUT, a new cost function called Weighted Sum of Absolute Difference (WSAD) is also proposed for further improvement. In our experiment, the classifier data size is reduced by 43{\%} and the computation cost is reduced by 15{\%} with same accuracy, compared to a conventional fixed LUT classifier.",
keywords = "Dynamic programming, Object detection, Signal partitioning, Wearable computing",
author = "Yuya Hanai and Tadahiro Kuroda",
year = "2009",
doi = "10.1109/ICIP.2009.5413645",
language = "English",
isbn = "9781424456543",
pages = "1225--1228",
booktitle = "Proceedings - International Conference on Image Processing, ICIP",
publisher = "IEEE Computer Society",

}

TY - GEN

T1 - Face detection through compact classifier using adaptive look-up-table

AU - Hanai, Yuya

AU - Kuroda, Tadahiro

PY - 2009

Y1 - 2009

N2 - Face detection has been well studied in terms of accuracy and speed. However, required memory size reduction is still poorly studied, which is becoming a critical issue as platforms for face detection go tiny. In this paper, we propose a novel compact weak classifier using Adaptive Look-Up-Table (ALUT) for face detection on resource-constrained devices such as wearable sensor nodes. ALUT gives good approximation of log-likelihood [3] with fewer data, thus enabling the drastic reduction of classifier data size, keeping high accuracy and low computation cost. To generate an optimal ALUT, a new cost function called Weighted Sum of Absolute Difference (WSAD) is also proposed for further improvement. In our experiment, the classifier data size is reduced by 43% and the computation cost is reduced by 15% with same accuracy, compared to a conventional fixed LUT classifier.

AB - Face detection has been well studied in terms of accuracy and speed. However, required memory size reduction is still poorly studied, which is becoming a critical issue as platforms for face detection go tiny. In this paper, we propose a novel compact weak classifier using Adaptive Look-Up-Table (ALUT) for face detection on resource-constrained devices such as wearable sensor nodes. ALUT gives good approximation of log-likelihood [3] with fewer data, thus enabling the drastic reduction of classifier data size, keeping high accuracy and low computation cost. To generate an optimal ALUT, a new cost function called Weighted Sum of Absolute Difference (WSAD) is also proposed for further improvement. In our experiment, the classifier data size is reduced by 43% and the computation cost is reduced by 15% with same accuracy, compared to a conventional fixed LUT classifier.

KW - Dynamic programming

KW - Object detection

KW - Signal partitioning

KW - Wearable computing

UR - http://www.scopus.com/inward/record.url?scp=77951964054&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77951964054&partnerID=8YFLogxK

U2 - 10.1109/ICIP.2009.5413645

DO - 10.1109/ICIP.2009.5413645

M3 - Conference contribution

AN - SCOPUS:77951964054

SN - 9781424456543

SP - 1225

EP - 1228

BT - Proceedings - International Conference on Image Processing, ICIP

PB - IEEE Computer Society

ER -