RECORD
Face Alignment using a 3D Deeply-initialized Ensemble of Regression Trees
Computer Vision and Pattern Recognition · Submitted 5 Feb 2019 · Published version
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim In the experiments performed, 3DDE improves the state-of-the-art in 300W, COFW, AFLW and WFLW data sets.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
Face alignment algorithms locate a set of landmark points in images of faces taken in unrestricted situations. State-of-the-art approaches typically fail or lose accuracy in the presence of occlusions, strong deformations, large pose variations and ambiguous configurations. In this paper we present 3DDE, a robust and efficient face alignment algorithm based on a coarse-to-fine cascade of ensembles of regression trees. It is initialized by robustly fitting a 3D face model to the probability maps produced by a convolutional neural network. With this initialization we address self-occlusions and large face rotations. Further, the regressor implicitly imposes a prior face shape on the solution, addressing occlusions and ambiguous face configurations. Its coarse-to-fine structure tackles the combinatorial explosion of parts deformation. In the experiments performed, 3DDE improves the state-of-the-art in 300W, COFW, AFLW and WFLW data sets. Finally, we perform cross-dataset experiments that reveal the existence of a significant data set bias in these benchmarks.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Face alignment algorithms locate a set of landmark points in images of faces taken in unrestricted situations. | background 1.00 | 0.07 | 0.02 | asserted 0.99 |
| s1 | State-of-the-art approaches typically fail or lose accuracy in the presence of occlusions, strong deformations, large pose variations and ambiguous configurations. | gap 0.61 | 0.07 | 0.03 | asserted 0.99 |
| s2 | In this paper we present 3DDE, a robust and efficient face alignment algorithm based on a coarse-to-fine cascade of ensembles of regression trees. | contribution 0.89 | 0.90 | 0.24 | asserted 1.00 |
| s3 | It is initialized by robustly fitting a 3D face model to the probability maps produced by a convolutional neural network. | method 0.98 | 0.86 | 0.04 | asserted 1.00 |
| s4 | With this initialization we address self-occlusions and large face rotations. | method 0.48 | 0.78 | 0.05 | asserted 1.00 |
| s5 | Further, the regressor implicitly imposes a prior face shape on the solution, addressing occlusions and ambiguous face configurations. | method 0.76 | 0.69 | 0.07 | asserted 1.00 |
| s6 | Its coarse-to-fine structure tackles the combinatorial explosion of parts deformation. | method 0.94 | 0.61 | 0.05 | asserted 1.00 |
| s7 | In the experiments performed, 3DDE improves the state-of-the-art in 300W, COFW, AFLW and WFLW data sets. | result 1.00 | 0.96 | 0.59 | asserted 1.00 |
| s8 | Finally, we perform cross-dataset experiments that reveal the existence of a significant data set bias in these benchmarks. | result 0.82 | 0.96 | 0.17 | asserted 0.99 |
What was examined, and what was not
- arXiv listing abstract v2
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
Face alignment algorithms locate a set of landmark points in images of faces taken in unrestricted situations. State-of-the-art approaches typically fail or lose accuracy in the presence of occlusions, strong deformations, large pose variations and ambiguous configurations. In this paper we present 3DDE, a robust and efficient face alignment algorithm based on a coarse-to-fine cascade of ensembles of regression trees. It is initialized by robustly fitting a 3D face model to the probability maps produced by a convolutional neural network. With this initialization we address self-occlusions and large face rotations. Further, the regressor implicitly imposes a prior face shape on the solution, addressing occlusions and ambiguous face configurations. Its coarse-to-fine structure tackles the combinatorial explosion of parts deformation. In the experiments performed, 3DDE improves the state-of-the-art in 300W, COFW, AFLW and WFLW data sets. Finally, we perform cross-dataset experiments that reveal the existence of a significant data set bias in these benchmarks.
Human review
How the literature uses it
1 arXiv paper cite this in 1 sentence
Papers citing it
Compares with it 1 paper
-
In<ref>, we compare our results with LBF [ref] , TCDCN [ref] , CFSS [ref] , MDM [ref] , RAR [ref] , DAN [ref] , TSR [ref] , SHN [ref] , LAB [ref] , DCFE [ref] , 3DDE [ref] , PCD-CNN [ref] , SAN [ref] , DeCaFA [ref] , AGCFN [ref] and ODN [ref] are also used in<ref>.
Multistage Model for Robust Face Alignment Using Deep Neural Networks · Experiment using 300-W dataset · 0.99
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.