کد مقاله کد نشریه سال انتشار مقاله انگلیسی نسخه تمام متن
526728 869216 2016 14 صفحه PDF دانلود رایگان
عنوان انگلیسی مقاله ISI
Real-time facial action unit intensity prediction with regularized metric learning *
ترجمه فارسی عنوان
پیش بینی شدت واحد فعلی صورت واقعی با یادگیری متریک منظم *
موضوعات مرتبط
مهندسی و علوم پایه مهندسی کامپیوتر چشم انداز کامپیوتر و تشخیص الگو
چکیده انگلیسی


• We present a framework for real-time Action Unit intensity prediction.
• We introduce a Lasso-regularized version of Metric Learning for Kernel Regression.
• We propose a new evaluation metric (r-AUC) designed for regression tasks.

The ability to automatically infer emotional states, engagement, depression or pain from nonverbal behavior has recently become of great interest in many research and industrial works. This will result in the emergence of a wide range of applications in robotics, biometrics, marketing and medicine. The Facial Action Coding System (FACS) proposed by Ekman features objective descriptions of facial movements, characterizing activations of facial muscles. Achieving an accurate intensity prediction of Action Units (AUs) has a significant impact on the prediction quality of more high-level information regarding human behavior (e.g. emotional states). Real-time AU intensity prediction, in many image-related machine learning tasks, is a high-dimensional problem. For solving this task, we propose adapting the Metric Learning for Kernel Regression (MLKR) framework focusing on overfitting issues induced in high-dimensional spaces. MLKR aims at estimating the optimal linear subspace for reducing the squared error of a Gaussian kernel regressor. We introduce Iterative Regularized Kernel Regression (IRKR), an iterative nonlinear feature selection method combined with a Lasso-regularized version of the original MLKR formulation that improves on the state-of-the-art results on several AU databases, ranging from prototypical to natural and wild data.

ناشر
Database: Elsevier - ScienceDirect (ساینس دایرکت)
Journal: Image and Vision Computing - Volume 52, August 2016, Pages 1–14
نویسندگان
, , ,