Feature selection for improved 3D facial expression recognition

Yurtkan, Kamil; Demirel, Hasan

doi:10.1016/j.patrec.2013.10.026

Cited by 39 publications

(11 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…For all the experiments, so far described in this paper, is has been assumed that the training set is representative of the possible different head poses in the test dataset. However, collecting the training data for all possible different head poses is not feasible in practice, also robustly estimating the head pose orientation from 2D data is still a challenge [3].…”

Section: Varied Head Posesmentioning

confidence: 99%

“…These features can be either hand-designed or learned from the training data. It is known that some features are more critical for analysing facial expressions than the others and the feature selection procedure can be applied to improve the performance [3], [4]. Indeed, extracting complex 2D or 3D features can improve the systems performance, but often requires more computational resources.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Is 2D Unlabeled Data Adequate for Recognizing Facial Expressions?

Tao

Matuszewski²

2016

IEEE Intell. Syst.

View full text Add to dashboard Cite

show abstract

Section: Varied Head Posesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Is 2D Unlabeled Data Adequate for Recognizing Facial Expressions?

Tao

Matuszewski²

2016

IEEE Intell. Syst.

View full text Add to dashboard Cite

show abstract

“…Although some of the methods achieved good performance, their performance can be improved further if two types of features were used. [23] 83.6% Geometric Rabiu et al [13] 92.2% Geometric Soyel et al [18] 91.3% Geometric Xioli et al [9] 90.2% Geometric Tekguc et al [19] 88.1% Geometric T.Yun [25] 85.39% Texture Lemaire et al [8] 78.13% Texture Yurtkan et al [26] Fig. 3.…”

Section: A Bu-3dfe Database 1) Datamentioning

confidence: 99%

“…Rabiu et al [13] has used the geometric data to obtain 16 feature distances based on the FACS principle, along with 27 Angles using maximum relevance minimum redundancy (mRMR) to reduce the features and then a Support Vector Machine (SVM) for classification. Yurtkan et al [26] recently propose a feature selection procedure for improved facial expression recognition utilizing 3-Dimensional (3D) geometrical facial feature point positions. It classifies expressions in six basic emotional categories which are anger, disgust, fear, happiness, sadness and surprise.…”

Section: Introductionmentioning

confidence: 99%

Automatic 3D facial expression recognition using geometric and textured feature fusion

Jan

Meng

2015

2015 11th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition (FG)

View full text Add to dashboard Cite

Abstract-3D facial expression recognition has gained more and more interests from affective computing society due to issues such as pose variations and illumination changes caused by 2D imaging having been eliminated. There are many applications that can benefit from this research, such as medical applications involving the detection of pain and psychological effects in patients, in human-computer interaction tasks that intelligent systems use in today's world. In this paper, we look into 3D Facial Expression Recognition, by investigating many feature extraction methods used on the 2D textured images and 3D geometric data, fusing the 2 domains to increase the overall performance. A One Vs All Multi-class SVM Classifier has been adopted to recognize the expressions Angry, Disgust, Fear, Happy, Neutral, Sad and Surprise from the BU-3DFE and Bosphorus databases. The proposed approach displays an increase in performance when the features are fused together.

show abstract

“…Bu bildiride daha önceki çalı malarımızda önerdi imiz yüz öznitelik noktalarının entropi de erlerine ba lı özgün öznitelik seçim yönteminden [1,2] ilham alınmı tır. Bu bildiride farklı olarak yüz temsili 3B yüz öznitelik uzaklıklarıyla yapılmı tır.…”

unclassified

Entropy driven feature selection for facial expression recognition based on 3-D facial feature distances

Yurtkan

Soyel

Demirel

2015

2015 23nd Signal Processing and Communications Applications Conference (SIU)

Self Cite

View full text Add to dashboard Cite

Özetçe-Bu belge, Yüz ifadeleri, insan-makine etkileşiminde (İME) insanın duygu durumu hakkında oldukça önemli bilgi içermesinden dolayı kilit bir noktadadır. Bu bildiride, insan yüz ifadelerini 6 ana sınıfa ayırabilmek için 3 boyutlu (3B) geometrik yüz öznitelikleri tabanlı yüz ifadeleri tanıma sistemine yönelik entropiye dayalı öznitelik seçimi sunulmaktadır. Yüz ifadeleri esnasında yüzdeki geometrik aktiviteleri ayırabilmek için daha önceki çalışmalarımızda önerdiğimiz entropiye dayalı öznitelik seçimi yüz üzerindeki 83 öznitelik noktalarının birbirlerine olan uzaklık bilgisi kullanılarak geliştirilmiştir. 3B yüz öznitelik uzaklıkları yüzdeki dönüşlere karşı daha gürbüz olduğu için kullanılan öznitelikler daha güvenilir bilgi taşımaktadır. Bildiride, daha önceki çalışmamızda yüz ifadeleri analizi için önerdiğimiz entropiye dayalı özgün öznitelik seçim yöntemi 3B öznitelik uzaklıklarına uygulanmıştır ve kişiden bağımsız olarak test edilmiştir. BU-3DFE veritabanı üzerinde test edilen system ümit verici sonuçlar elde edilmiştir. Anahtar Kelimeler-Yüz ifadeleri analizi, yüz ifadeleri tanıma, öznitelik seçimi, yüz biometrisi, entropi.Abstract-Facial expressions contain a lot of information about the feelings of a human. It plays an important role in humancomputer interaction. In this paper, entropy based feature selection method applied to 3D facial feature distances is presented for a facial expression recognition system classifying the expressions into 6 basic classes based on 3-Dimensional (3D) face geometry. Our previous work on entropy based feature selection has been improved by employing 3D feature distances between the 83 points on the face as facial features. 3D distances are more robust to rotations of the face and involve more accurate information than 3D feature positions that are used in our previous work. Entropy is applied in order to rank the feature distances for feature selection. The system is tested on BU-3DFE database in person independent manner and provides encouraging recognition rates.

show abstract

Feature selection for improved 3D facial expression recognition

Cited by 39 publications

References 8 publications

Is 2D Unlabeled Data Adequate for Recognizing Facial Expressions?

Is 2D Unlabeled Data Adequate for Recognizing Facial Expressions?

Automatic 3D facial expression recognition using geometric and textured feature fusion

Entropy driven feature selection for facial expression recognition based on 3-D facial feature distances

Contact Info

Product

Resources

About