Bio-inspired hierarchical framework for multi-view face detection and pose estimation

N McCarroll, Ammar Belatreche, Jim Harkin, Yuhua Li

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

3 Citations (Scopus)

Abstract

Face detection is one of the most active research areas in computer vision. Despite the well documented success of classical machine learning techniques in controlled situations, face detection in completely uncontrolled settings remains a difficult task. Recent progress with bio-inspired approaches have addressed challenging areas of invariance including scale, occlusion and illumination issues but there remains a lack of concentrated effort into truly multi-view detection of faces in different poses and orientations. This paper introduces a novel strategy to address this through the enhanced implementation of a hierarchical bio-inspired HMAX framework using spiking neurons that implements feature extraction with unsupervised STDP. A multiple trial training scheme is introduced to train separate pools of neurons on different face poses. The trained neurons are then processed by an additional STDP mechanism to generate a streamlined repository of broadly tuned multi-view neurons. Experimental results demonstrate that the new system achieves robust invariant detection of in-plane and out-of-plane rotated faces with single face per image datasets. In addition, extending the multi-view system by introducing lateral inhibition between merged pools of multi-view face detecting neurons, results in a single model that is able to achieve simultaneous face detection and accurate face pose estimation.
LanguageEnglish
Title of host publicationUnknown Host Publication
Number of pages6
Publication statusPublished - 12 Jul 2015
EventInternational Joint Conference on Neural Networks (IJCNN) - Ireland
Duration: 12 Jul 2015 → …

Conference

ConferenceInternational Joint Conference on Neural Networks (IJCNN)
Period12/07/15 → …

Fingerprint

Face recognition
Neurons
Invariance
Computer vision
Learning systems
Feature extraction
Lighting

Keywords

  • Multi-view face detection
  • pose estimation
  • Spiking Neural Networks
  • STDP
  • Hierarchical object detection

Cite this

McCarroll, N., Belatreche, A., Harkin, J., & Li, Y. (2015). Bio-inspired hierarchical framework for multi-view face detection and pose estimation. In Unknown Host Publication
McCarroll, N ; Belatreche, Ammar ; Harkin, Jim ; Li, Yuhua. / Bio-inspired hierarchical framework for multi-view face detection and pose estimation. Unknown Host Publication. 2015.
@inproceedings{f9f9528d344a492dbeb7ada391f305e7,
title = "Bio-inspired hierarchical framework for multi-view face detection and pose estimation",
abstract = "Face detection is one of the most active research areas in computer vision. Despite the well documented success of classical machine learning techniques in controlled situations, face detection in completely uncontrolled settings remains a difficult task. Recent progress with bio-inspired approaches have addressed challenging areas of invariance including scale, occlusion and illumination issues but there remains a lack of concentrated effort into truly multi-view detection of faces in different poses and orientations. This paper introduces a novel strategy to address this through the enhanced implementation of a hierarchical bio-inspired HMAX framework using spiking neurons that implements feature extraction with unsupervised STDP. A multiple trial training scheme is introduced to train separate pools of neurons on different face poses. The trained neurons are then processed by an additional STDP mechanism to generate a streamlined repository of broadly tuned multi-view neurons. Experimental results demonstrate that the new system achieves robust invariant detection of in-plane and out-of-plane rotated faces with single face per image datasets. In addition, extending the multi-view system by introducing lateral inhibition between merged pools of multi-view face detecting neurons, results in a single model that is able to achieve simultaneous face detection and accurate face pose estimation.",
keywords = "Multi-view face detection, pose estimation, Spiking Neural Networks, STDP, Hierarchical object detection",
author = "N McCarroll and Ammar Belatreche and Jim Harkin and Yuhua Li",
year = "2015",
month = "7",
day = "12",
language = "English",
booktitle = "Unknown Host Publication",

}

McCarroll, N, Belatreche, A, Harkin, J & Li, Y 2015, Bio-inspired hierarchical framework for multi-view face detection and pose estimation. in Unknown Host Publication. International Joint Conference on Neural Networks (IJCNN), 12/07/15.

Bio-inspired hierarchical framework for multi-view face detection and pose estimation. / McCarroll, N; Belatreche, Ammar; Harkin, Jim; Li, Yuhua.

Unknown Host Publication. 2015.

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

TY - GEN

T1 - Bio-inspired hierarchical framework for multi-view face detection and pose estimation

AU - McCarroll, N

AU - Belatreche, Ammar

AU - Harkin, Jim

AU - Li, Yuhua

PY - 2015/7/12

Y1 - 2015/7/12

N2 - Face detection is one of the most active research areas in computer vision. Despite the well documented success of classical machine learning techniques in controlled situations, face detection in completely uncontrolled settings remains a difficult task. Recent progress with bio-inspired approaches have addressed challenging areas of invariance including scale, occlusion and illumination issues but there remains a lack of concentrated effort into truly multi-view detection of faces in different poses and orientations. This paper introduces a novel strategy to address this through the enhanced implementation of a hierarchical bio-inspired HMAX framework using spiking neurons that implements feature extraction with unsupervised STDP. A multiple trial training scheme is introduced to train separate pools of neurons on different face poses. The trained neurons are then processed by an additional STDP mechanism to generate a streamlined repository of broadly tuned multi-view neurons. Experimental results demonstrate that the new system achieves robust invariant detection of in-plane and out-of-plane rotated faces with single face per image datasets. In addition, extending the multi-view system by introducing lateral inhibition between merged pools of multi-view face detecting neurons, results in a single model that is able to achieve simultaneous face detection and accurate face pose estimation.

AB - Face detection is one of the most active research areas in computer vision. Despite the well documented success of classical machine learning techniques in controlled situations, face detection in completely uncontrolled settings remains a difficult task. Recent progress with bio-inspired approaches have addressed challenging areas of invariance including scale, occlusion and illumination issues but there remains a lack of concentrated effort into truly multi-view detection of faces in different poses and orientations. This paper introduces a novel strategy to address this through the enhanced implementation of a hierarchical bio-inspired HMAX framework using spiking neurons that implements feature extraction with unsupervised STDP. A multiple trial training scheme is introduced to train separate pools of neurons on different face poses. The trained neurons are then processed by an additional STDP mechanism to generate a streamlined repository of broadly tuned multi-view neurons. Experimental results demonstrate that the new system achieves robust invariant detection of in-plane and out-of-plane rotated faces with single face per image datasets. In addition, extending the multi-view system by introducing lateral inhibition between merged pools of multi-view face detecting neurons, results in a single model that is able to achieve simultaneous face detection and accurate face pose estimation.

KW - Multi-view face detection

KW - pose estimation

KW - Spiking Neural Networks

KW - STDP

KW - Hierarchical object detection

M3 - Conference contribution

BT - Unknown Host Publication

ER -

McCarroll N, Belatreche A, Harkin J, Li Y. Bio-inspired hierarchical framework for multi-view face detection and pose estimation. In Unknown Host Publication. 2015