Fernando De la Torre

Supervised Descent Method and Its Applications to Face Alignment

Many computer vision problems (e.g., camera calibration, image alignment, structure from motion) are solved through a nonlinear optimization method. It is generally accepted that 2nd order descent methods are the most robust, fast and reliable approaches for nonlinear optimization of a general smooth function. However, in the context of computer vision, 2nd order descent methods have two main drawbacks: (1) The function might not be analytically differentiable and numerical approximations are impractical. (2) The Hessian might be large and not positive definite. To address these issues, this paper proposes a Supervised Descent Method (SDM) for minimizing a Non-linear Least Squares (NLS) function. During training, the SDM learns a sequence of descent directions that minimizes the mean of NLS functions sampled at different points. In testing, SDM minimizes the NLS objective using the learned descent directions without computing the Jacobian nor the Hessian. We illustrate the benefits of our approach in synthetic and real examples, and show how SDM achieves state-of-the-art performance in the problem of facial feature detection. The code is available at www.humansensing.cs. cmu.edu/intraface.

A Framework for Robust Subspace Learning

Many computer vision, signal processing and statistical problems can be posed as problems of learning low dimensional linear or multi-linear models. These models have been widely used for the representation of shape, appearance, motion, etc., in computer vision applications. Methods for learning linear models can be seen as a special case of subspace fitting. One draw-back of previous learning methods is that they are based on least squares estimation techniques and hence fail to account for “outliers” which are common in realistic training sets. We review previous approaches for making linear learning methods robust to outliers and present a new method that uses an intra-sample outlier process to account for pixel outliers. We develop the theory of Robust Subspace Learning (RSL) for linear models within a continuous optimization framework based on robust M-estimation. The framework applies to a variety of linear learning problems in computer vision including eigen-analysis and structure from motion. Several synthetic and natural examples are used to develop and illustrate the theory and applications of robust subspace learning in computer vision.

Robust regression

Discriminative methods (e.g., kernel regression, SVM) have been extensively used to solve problems such as object recognition, image alignment and pose estimation from images. These methods typically map image features (X) to continuous (e.g., pose) or discrete (e.g., object category) values. A major drawback of existing discriminative methods is that samples are directly projected onto a subspace and hence fail to account for outliers common in realistic training sets due to occlusion, specular reflections or noise. It is important to notice that existing discriminative approaches assume the input variables X to be noise free. Thus, discriminative methods experience significant performance degradation when gross outliers are present. Despite its obvious importance, the problem of robust discriminative learning has been relatively unexplored in computer vision. This paper develops the theory of robust regression (RR) and presents an effective convex approach that uses recent advances on rank minimization. The framework applies to a variety of problems in computer vision including robust linear discriminant analysis, regression with missing data, and multi-label classification. Several synthetic and real examples with applications to head pose estimation from images, image and video classification and facial attribute classification with missing data are used to illustrate the benefits of RR.

Joint segmentation and classification of human actions in video

Automatic video segmentation and action recognition has been a long-standing problem in computer vision. Much work in the literature treats video segmentation and action recognition as two independent problems; while segmentation is often done without a temporal model of the activity, action recognition is usually performed on pre-segmented clips. In this paper we propose a novel method that avoids the limitations of the above approaches by jointly performing video segmentation and action recognition. Unlike standard approaches based on extensions of dynamic Bayesian networks, our method is based on a discriminative temporal extension of the spatial bag-of-words model that has been very popular in object recognition. The classification is performed robustly within a multi-class SVM framework whereas the inference over the segments is done efficiently with dynamic programming. Experimental results on honeybee, Weizmann, and Hollywood datasets illustrate the benefits of our approach compared to state-of-the-art methods.

Detecting depression from facial actions and vocal prosody

Current methods of assessing psychopathology depend almost entirely on verbal report (clinical interview or questionnaire) of patients, their family, or caregivers. They lack systematic and efficient ways of incorporating behavioral observations that are strong indicators of psychological disorder, much of which may occur outside the awareness of either individual. We compared clinical diagnosis of major depression with automatically measured facial actions and vocal prosody in patients undergoing treatment for depression. Manual FACS coding, active appearance modeling (AAM) and pitch extraction were used to measure facial and vocal expression. Classifiers using leave-one-out validation were SVM for FACS and for AAM and logistic regression for voice. Both face and voice demonstrated moderate concurrent validity with depression. Accuracy in detecting depression was 88% for manual FACS and 79% for AAM. Accuracy for vocal prosody was 79%. These findings suggest the feasibility of automatic detection of depression, raise new issues in automated facial image analysis and machine learning, and have exciting implications for clinical theory and practice.

Selective Transfer Machine for Personalized Facial Action Unit Detection

Automatic facial action unit (AFA) detection from video is a long-standing problem in facial expression analysis. Most approaches emphasize choices of features and classifiers. They neglect individual differences in target persons. People vary markedly in facial morphology (e.g., heavy versus delicate brows, smooth versus deeply etched wrinkles) and behavior. Individual differences can dramatically influence how well generic classifiers generalize to previously unseen persons. While a possible solution would be to train person-specific classifiers, that often is neither feasible nor theoretically compelling. The alternative that we propose is to personalize a generic classifier in an unsupervised manner (no additional labels for the test subjects are required). We introduce a transductive learning method, which we refer to Selective Transfer Machine (STM), to personalize a generic classifier by attenuating person-specific biases. STM achieves this effect by simultaneously learning a classifier and re-weighting the training samples that are most relevant to the test subject. To evaluate the effectiveness of STM, we compared STM to generic classifiers and to cross-domain learning methods in three major databases: CK+, GEMEP-FERA and RU-FACS. STM outperformed generic classifiers in all.

Max-Margin Early Event Detectors

The need for early detection of temporal events from sequential data arises in a wide spectrum of applications ranging from human-robot interaction to video security. While temporal event detection has been extensively studied, early detection is a relatively unexplored problem. This paper proposes a maximum-margin framework for training temporal event detectors to recognize partial events, enabling early detection. Our method is based on Structured Output SVM, but extends it to accommodate sequential data. Experiments on datasets of varying complexity, for detecting facial expressions, hand gestures, and human activities, demonstrate the benefits of our approach.

Optimal feature selection for support vector machines

Selecting relevant features for support vector machine (SVM) classifiers is important for a variety of reasons such as generalization performance, computational efficiency, and feature interpretability. Traditional SVM approaches to feature selection typically extract features and learn SVM parameters independently. Independently performing these two steps might result in a loss of information related to the classification process. This paper proposes a convex energy-based framework to jointly perform feature selection and SVM parameter learning for linear and non-linear kernels. Experiments on various databases show significant reduction of features used while maintaining classification performance.

Aligned Cluster Analysis for temporal segmentation of human motion

Temporal segmentation of human motion into actions is a crucial step for understanding and building computational models of human motion. Several issues contribute to the challenge of this task. These include the large variability in the temporal scale and periodicity of human actions, as well as the exponential nature of all possible movement combinations. We formulate the temporal segmentation problem as an extension of standard clustering algorithms. In particular, this paper proposes aligned cluster analysis (ACA), a robust method to temporally segment streams of motion capture data into actions. ACA extends standard kernel k-means clustering in two ways: (1) the cluster means contain a variable number of features, and (2) a dynamic time warping (DTW) kernel is used to achieve temporal invariance. Experimental results, reported on synthetic data and the Carnegie Mellon Motion Capture database, demonstrate its effectiveness.

Temporal segmentation and activity classification from first-person sensing

Temporal segmentation of human motion into actions is central to the understanding and building of computational models of human motion and activity recognition. Several issues contribute to the challenge of temporal segmentation and classification of human motion. These include the large variability in the temporal scale and periodicity of human actions, the complexity of representing articulated motion, and the exponential nature of all possible movement combinations. We provide initial results from investigating two distinct problems -classification of the overall task being performed, and the more difficult problem of classifying individual frames over time into specific actions. We explore first-person sensing through a wearable camera and inertial measurement units (IMUs) for temporally segmenting human motion into actions and performing activity classification in the context of cooking and recipe preparation in a natural environment. We present baseline results for supervised and unsupervised temporal segmentation, and recipe recognition in the CMU-multimodal activity database (CMU-MMAC).