|
Descripción
|
|
|---|---|
| In this paper we present DCFE, a real-time facial landmarkregression method based on a coarse-to-fine Ensemble of Regression Trees(ERT). We use a simple Convolutional Neural Network (CNN) to gen-erate probability maps of landmarks location. These are further refinedwith the ERT regressor, which is initialized by fitting a 3D face modelto the landmark maps. The coarse-to-fine structure of the ERT letsusaddress the combinatorial explosion of parts deformation. Withthe 3Dmodel we also tackle other key problems such as robust regressor initial-ization, self occlusions, and simultaneous frontal and profileface analysis.In the experiments DCFE achieves the best reported result in AFLW,COFW, and 300W private and common public data sets. | |
|
Internacional
|
Si |
|
Nombre congreso
|
Proceedings European Conference on Computer Vision, ECCV |
|
Tipo de participación
|
960 |
|
Lugar del congreso
|
|
|
Revisores
|
Si |
|
ISBN o ISSN
|
978-3-030-01263-2 |
|
DOI
|
|
|
Fecha inicio congreso
|
08/09/2018 |
|
Fecha fin congreso
|
14/09/2018 |
|
Desde la página
|
609 |
|
Hasta la página
|
624 |
|
Título de las actas
|
A Deeply-initialized Coarse-to-fine Ensemble of Regression Trees for Face Alignment |