In many applications, a face recognition model learned on a source domain but applied to a novel target domain degenerates even significantly due to the mismatch between the two domains. Aiming at learning a better face recognition model for the target domain, this paper proposes a simple but effective domain adaptation approach that transfers the supervision knowledge from a labeled source domain to the unlabeled target domain. Our basic idea is to convert the source domain images to target domain (termed as targetize the source domain hereinafter), and at the same time keep its supervision information. For this purpose, each source domain image is simply represented as a linear combination of sparse target domain neighbors in the image space, with the combination coefficients however learnt in a common subspace. The principle behind this strategy is that, the common knowledge is only favorable for accurate cross-domain reconstruction, but for the classification in the target domain, the specific knowledge of the target domain is also essential and thus should be mostly preserved (through targetization in the image space in this work). To discover the common knowledge, specifically, a common subspace is learnt, in which the structures of both domains are preserved and meanwhile the disparity of source and target domains is reduced. The proposed method is extensively evaluated under three face recognition scenarios, i.e., domain adaptation across view angle, domain adaptation across ethnicity and domain adaptation across imaging condition. The experimental results illustrate the superiority of our method over those competitive ones.

Face recognition on large-scale video in the wild is becoming increasingly important due to the ubiquity of video data captured by surveillance cameras, handheld devices, Internet uploads, and other sources. By treating each video as one image set, set-based methods recently have made great success in the field of video-based face recognition. In the wild world, videos often contain extremely complex data variations and thus pose a big challenge of set modeling for set-based methods. In this paper, we propose a novel Hybrid Euclidean-and-Riemannian Metric Learning (HERML) method to fuse multiple statistics of image set. Specifically, we represent each image set simultaneously by mean, covariance matrix and Gaussian distribution, which generally complement each other in the aspect of set modeling. However, it is not trivial to fuse them since mean, covariance matrix and Gaussian model typically lie in multiple heterogeneous spaces equipped with Euclidean or Riemannian metric. Therefore, we first implicitly map the original statistics into high dimensional Hilbert spaces by exploiting Euclidean and Riemannian kernels. With a LogDet divergence based objective function, the hybrid kernels are then fused by our hybrid metric learning framework, which can efficiently perform the fusing procedure on large-scale videos. The proposed method is evaluated on four public and challenging large-scale video face datasets. Extensive experimental results demonstrate that our method has a clear superiority over the state-of-the-art set-based methods for large-scale video-based face recognition.

Video-based spontaneous expression recognition is a challenging task due to the large inter-personal variations of both the expressing manners and the executing rates for the same expression category. One of the key is to explore robust representation method which can effectively capture the facial variations as well as alleviate the influence of personalities. In this paper, we propose to learn a kind of typical patterns that can be commonly shared by different subjects when performing expressions, namely “prototypes”. Specifically, we first apply a statistical model (i.e. linear subspace) on facial regions to generate the specific expression patterns for each video. Then a clustering algorithm is employed on all these expression patterns and the cluster means are regarded as the “prototypes”. Accordingly, we further design “simile” features to measure the similarities of personal specific patterns to our learned “prototypes”. Both techniques are conducted on Grassmann manifold, which can enrich the feature encoding manners and better reveal the data structure by introducing intrinsic geodesics. Extensive experiments are conducted on both posed and spontaneous expression databases. All results show that our method outperforms the state-of-the-art and also possesses good transferable ability under cross-database scenario.

Facial expression is a temporally dynamic event which can be decomposed into a set of muscle motions occurring in different facial regions over various time intervals. For dynamic expression recognition, two key issues, temporal alignment and semantics-aware dynamic representation, must be taken into account. In this paper, we attempt to solve both problems via manifold modeling of videos based on a novel mid-level representation, i.e., expressionlet. Specifically, our method contains three key stages: 1) each expression video clip is characterized as a spatial-temporal manifold (STM) formed by dense low-level features; 2) a universal manifold model (UMM) is learned over all low-level features and represented as a set of local modes to statistically unify all the STMs; and 3) the local modes on each STM can be instantiated by fitting to the UMM, and the corresponding expressionlet is constructed by modeling the variations in each local mode. With the above strategy, expression videos are naturally aligned both spatially and temporally. To enhance the discriminative power, the expressionlet-based STM representation is further processed with discriminant embedding. Our method is evaluated on four public expression databases, CK+, MMI, Oulu-CASIA, and FERA. In all cases, our method outperforms the known state of the art by a large margin.

We address the problem of face video retrieval in TV-series, which searches video clips based on the presence of specific character, given one face track of his/her. This is tremendously challenging because on one hand, faces in TV-series are captured in largely uncontrolled conditions with complex appearance variations, and on the other hand, retrieval task typically needs efficient representation with low time and space complexity. To handle this problem, we propose a compact and discriminative representation for the huge body of video data, named compact video code (CVC). Our method first models the face track by its sample (i.e., frame) covariance matrix to capture the video data variations in a statistical manner. To incorporate discriminative information and obtain more compact video signature suitable for retrieval, the high-dimensional covariance representation is further encoded as a much lower dimensional binary vector, which finally yields the proposed CVC. Specifically, each bit of the code, i.e., each dimension of the binary vector, is produced via supervised learning in a max margin framework, which aims to make a balance between the discriminability and stability of the code. Besides, we further extend the descriptive granularity of covariance matrix from traditional pixel-level to more general patch-level, and proceed to propose a novel hierarchical video representation named spatial pyramid covariance along with a fast calculation method. Face retrieval experiments on two challenging TV-series video databases, i.e., the Big Bang Theory and Prison Break, demonstrate the competitiveness of the proposed CVC over the state-of-the-art retrieval methods. In addition, as a general video matching algorithm, CVC is also evaluated in traditional video face recognition task on a standard Internet database, i.e., YouTube Celebrities, showing its quite promising performance by using an extremely compact code with only 128 bits.