Gradient Shape Model
- PDF / 2,170,885 Bytes
- 21 Pages / 595.276 x 790.866 pts Page_size
- 24 Downloads / 208 Views
Gradient Shape Model Pedro Martins1
· João F. Henriques2 · Jorge Batista1,3
Received: 19 May 2018 / Accepted: 12 May 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract For years, the so-called Constrained Local Model (CLM) and its variants have been the gold standard in face alignment tasks. The CLM combines an ensemble of local feature detectors whose locations are regularized by a shape model. Fitting such a model typically consists of an exhaustive local search using the detectors and a global optimization that finds the CLM’s parameters that jointly maximize all the responses. However, one major drawback of CLMs is the inefficiency of the local search, which relies on a large amount of expensive convolutions. This paper introduces the Gradient Shape Model (GSM), a novel approach that addresses this limitation. We are able to align a similar CLM model without the need for any convolutions at all. We also use true analytical gradient and Hessian matrices, which are easy to compute, instead of their approximations. Our formulation is very general, allowing an optional 3D shape term to be seamlessly included. Additionally, we expand the GSM formulation through a cascade regression framework. This revised technique allows a substantially reduction in the complexity/dimensionality of the data term, making it possible to compute a denser, more accurate, regression step per cascade level. Experiments in several standard datasets show that our proposed models perform faster than state-of-the-art CLMs and better than recent cascade regression approaches. Keywords Facial landmark localization · Face alignment · Constrained local model · CLM.
1 Introduction Nonrigid face registration, sometimes known as facial alignment, plays a fundamental role in many computer vision tasks. Typical applications include visual tracking, recognition (identity and facial expression), biometric security, video compression, head pose estimation and many others. In general, the main goal of face registration consists of locating, with accuracy, the semantic structural facial landmarks (fiducial points) such as eyes, nose, mouth, chin, eyebrows, Communicated by Ming-Hsuan Yang.
B
Pedro Martins [email protected] João F. Henriques [email protected] Jorge Batista [email protected]
1
Institute of Systems and Robotics, University of Coimbra, Coimbra, Portugal
2
Visual Geometry Group, University of Oxford, Oxford, UK
3
Department of Electrical and Computer Engineering, University of Coimbra, Coimbra, Portugal
etc. Although this problem has been studied for years, it still is challenging to locate and consistently track subjects with previously unseen appearances and under unconstrained acquisition conditions (p.e. changes in pose, lighting, occlusion, resolution and focus). Much has been done in the past, but since the introduction of the Active Shape Model (ASM) (Cootes et al. 1995) and shortly after with the Active Appearance Model (AAM) (Cootes et al. 2001; Matthews and Baker 2004; Alabort-iMedi
Data Loading...