Abstract
independently on the two features, are stacked and used with a Support Vector Machine (SVM). Experimental results show that the late fusion strategy gives the best results and the transition frames give better identification results as compared to the peak emotion frames.
Original language | English |
---|---|
Title of host publication | 13th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2018) |
Number of pages | 8 |
Publisher | IEEE |
Publication date | 2018 |
Pages | 621-628 |
Article number | 8373891 |
ISBN (Print) | 978-1-5386-2336-7 |
ISBN (Electronic) | 978-1-5386-2335-0 |
DOIs | |
Publication status | Published - 2018 |
Event | IEEE Conf. on Automatic Face and Gesture Recognition Workshops - X'ian, China Duration: 15 May 2018 → 19 May 2018 https://fg2018.cse.sc.edu |
Conference
Conference | IEEE Conf. on Automatic Face and Gesture Recognition Workshops |
---|---|
Country | China |
City | X'ian |
Period | 15/05/2018 → 19/05/2018 |
Internet address |
Fingerprint
Keywords
- Facial expression
- biometric
- database
- benchmark
- Deep Learning
- CNN
- LSTM
- Multimodal
- Spatio-temporal
- SVM
Cite this
}
Changes in Facial Expression as Biometric : A Database and Benchmarks of Identification. / Haamer, Rain Eric; Kulkarni, Kaustubh; Imanpour, Nasrin; Haque, Mohammad Ahsanul; Avots, Egils; Breisch, Michelle; Nasrollahi, Kamal; Guerrero, Sergio Escalera; Ozcinar, Cagri; Baro, Xavier; Naghsh-Nilchi, Ahmad Reza; Moeslund, Thomas B.; Anbarjafari, Gholamreza .
13th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2018). IEEE, 2018. p. 621-628 8373891.Research output: Contribution to book/anthology/report/conference proceeding › Article in proceeding › Research › peer-review
TY - GEN
T1 - Changes in Facial Expression as Biometric
T2 - A Database and Benchmarks of Identification
AU - Haamer, Rain Eric
AU - Kulkarni, Kaustubh
AU - Imanpour, Nasrin
AU - Haque, Mohammad Ahsanul
AU - Avots, Egils
AU - Breisch, Michelle
AU - Nasrollahi, Kamal
AU - Guerrero, Sergio Escalera
AU - Ozcinar, Cagri
AU - Baro, Xavier
AU - Naghsh-Nilchi, Ahmad Reza
AU - Moeslund, Thomas B.
AU - Anbarjafari, Gholamreza
PY - 2018
Y1 - 2018
N2 - Facial dynamics can be considered as unique signatures for discrimination between people. These have started to become important topic since many devices have the possibility of unlocking using face recognition or verification. In this work, we evaluate the efficacy of the transition frames of video in emotion as compared to the peak emotion frames for identification. For experiments with transition frames we extract features from each frame of the video from a fine-tuned VGG-Face Convolutional Neural Network (CNN) and geometric features from facial landmark points. To model the temporal context of the transition frames we train a Long-Short Term Memory (LSTM) on the geometric and the CNN features. Furthermore, we employ two fusion strategies: first, an early fusion, in which the geometric and the CNN features are stacked and fed to the LSTM. Second, a late fusion, in which the prediction of the LSTMs, trainedindependently on the two features, are stacked and used with a Support Vector Machine (SVM). Experimental results show that the late fusion strategy gives the best results and the transition frames give better identification results as compared to the peak emotion frames.
AB - Facial dynamics can be considered as unique signatures for discrimination between people. These have started to become important topic since many devices have the possibility of unlocking using face recognition or verification. In this work, we evaluate the efficacy of the transition frames of video in emotion as compared to the peak emotion frames for identification. For experiments with transition frames we extract features from each frame of the video from a fine-tuned VGG-Face Convolutional Neural Network (CNN) and geometric features from facial landmark points. To model the temporal context of the transition frames we train a Long-Short Term Memory (LSTM) on the geometric and the CNN features. Furthermore, we employ two fusion strategies: first, an early fusion, in which the geometric and the CNN features are stacked and fed to the LSTM. Second, a late fusion, in which the prediction of the LSTMs, trainedindependently on the two features, are stacked and used with a Support Vector Machine (SVM). Experimental results show that the late fusion strategy gives the best results and the transition frames give better identification results as compared to the peak emotion frames.
KW - Facial expression
KW - biometric
KW - database
KW - benchmark
KW - Deep Learning
KW - CNN
KW - LSTM
KW - Multimodal
KW - Spatio-temporal
KW - SVM
U2 - 10.1109/FG.2018.00098
DO - 10.1109/FG.2018.00098
M3 - Article in proceeding
SN - 978-1-5386-2336-7
SP - 621
EP - 628
BT - 13th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2018)
PB - IEEE
ER -