Facial expression transfer method based on frequency analysis

Wei, Wei; Tian, Chunna; Maybank, Stephen J.; Zhang, Yanning

doi:10.1016/j.patcog.2015.08.004

Cited by 15 publications

(2 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The general performance of the traditional classifier trained with a limited number of labelled data may not be satisfactory, while the manual annotation of abundant training data for various tasks costs too much manpower. Fortunately, transfer learning [25,26,27] is an effective technique to improve the performance of the classifier in the target domain given only the annotated data in the source domain, which greatly reduces the labelling cost. Transfer learning also refers to unsupervised domain adaptation [28,29,30], and it can adapt the features from the labelled source domain to the unlabelled target domain.…”

Section: Transfer Learningmentioning

confidence: 99%

Multi-label learning based deep transfer neural network for facial attribute classification

Zhuang

Yan

Chen

et al. 2018

Pattern Recognition

View full text Add to dashboard Cite

Deep Neural Network (DNN) has recently achieved outstanding performance in a variety of computer vision tasks, including facial attribute classification. The great success of classifying facial attributes with DNN often relies on a massive amount of labelled data. However, in real-world applications, labelled data are only provided for some commonly used attributes (such as age, gender); whereas, unlabelled data are available for other attributes (such as attraction, hairline). To address the above problem, we propose a novel deep transfer neural network method based on multi-label learning for facial attribute classification, termed FMTNet, which consists of three sub-networks: the Face detection Network (FNet), the Multi-label learning Network (MNet) and the Transfer learning Network (TNet). Firstly, based on the Faster Region-based Convolutional Neural Network (Faster R-CNN), FNet is finetuned for face detection. Then, MNet is fine-tuned by FNet to predict multiple attributes with labelled data, where an effective loss weight scheme is developed to explicitly exploit the correlation between facial attributes based on attribute grouping. Finally, based on MNet, TNet is trained by taking advantage of unsupervised domain adaptation for unlabelled facial attribute classification. The three sub-networks are tightly coupled to perform effective facial attribute classification. A distinguishing characteristic of the proposed FMTNet method is that the three sub-networks (FNet, MNet and TNet) are constructed in a similar network structure. Extensive experimental results on challenging face datasets demonstrate the effectiveness of our proposed method compared with several state-of-the-art methods.

show abstract

Section: Transfer Learningmentioning

confidence: 99%

Multi-label learning based deep transfer neural network for facial attribute classification

Zhuang

Yan

Chen

et al. 2018

Pattern Recognition

View full text Add to dashboard Cite

show abstract

“…The high quality of feature extraction can only ensure the success of time-driven pattern classi¯cation. Until now, several feature extraction methods, such as principal kernel principal component analysis (PCA), 25 independent component analysis (ICA), 12 dynamic time warping, 21 PCA, 24 arti¯cial neural network (ANN), 3 and wavelet transforms (WT) 8 have been widely used. Especially, the WT can extract rich problem-speci¯c information from sensor signals.…”

Section: Introductionmentioning

confidence: 99%

WSN-Based Vibration Characteristic Research for Various Railway Track Structures for Pattern Classification

Yue

Wang

et al. 2016

Int. J. Patt. Recogn. Artif. Intell.

View full text Add to dashboard Cite

To e®ectively study vibration characteristics of tracks under di®erent track structures, wavelet transforms of the vibration data are used for pattern classi¯cation of vibration feature. First, acceleration data of the track are collected with running speed of 150 km/h at 26 positions respectively on a slab tangent track, ballast tangent track and ballast curve track by a wireless sensor network (WSN). Then they are analyzed using the power spectral densities (PSDs) and wavelet-based energy spectrum analysis. The paper elaborates on the reasons for the di®erences of vibration energy and excitation frequencies due to the mechanism of di®erent frequency bands and the corresponding track structures. Based on these, the instantaneous frequencies, vibration energies and durations in the low, medium, and high frequency bands are selected as the features for three track structures. A function curve representing the features is proposed to detect the abnormal track structure by a correlation analysis. Finally, the proposed method of pattern classi¯cation has been validated by experimental testings.

show abstract