Multi-task convolution network for face alignment

Sun, Yang; Zhang, Xuan; Li, Chongrong

doi:10.1088/1742-6596/887/1/012079

Cited by 4 publications

(5 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We compare our approach with various state-of-the-art methods on the Common and Challenging subsets of 300W. These methods contain cascade regression [5], [6], [8], [9], [18] and CNN-based architectures [13], [32]- [34], [37]- [39], [47]. As shown in Table 1, our result on Common subset comes as the third best behind Deep Reg and DR.…”

Section: ) Comparison On 300wmentioning

confidence: 97%

“…There are two popular manners to learn a regression mapping between facial landmarks and image appearance, one of which is based on Deep Convolutional Neural Network. Except two aforementioned pioneering works [13], [32], Liu et al [37] introduced a 3D face model for face image and trained a CNN to fit it. Liu et al [35] used a multitude of facial feature detectors to distinguish semantic regions and process these regions with small sized neural networks that exploit both the structural and sharable information.…”

Section: Related Workmentioning

confidence: 99%

“…Liu et al [35] used a multitude of facial feature detectors to distinguish semantic regions and process these regions with small sized neural networks that exploit both the structural and sharable information. Sun et al [32] introduced an auxiliary task of pose estimation for face alignment and proposed a multi-task deep CNN. Jourabloo et al [33] stacked a series of visualization blocks to construct a new CNN architecture which is considered as a cascade of shallow CNNs.…”

Section: Related Workmentioning

confidence: 99%

“…We compare our experiment result on Helen dataset with the recent methods of cascade regressions [8], [11], [18] and recently CNN-based methods [13], [32], [38], [39]. Since the pupils are not annotated in the HELEN dataset, we use the distance between the centroids on two eyes to normalize the deviation from ground truth [6] instead.…”

Section: ) Comparison On Helenmentioning

confidence: 99%

“…Deep network and cascade regression are two ways to learn such a regression function, and both them have achieved success in face alignment [40]. Deep network methods [13], [32], [33], [35], [37]- [39] are presented in many literatures for localizing the facial landmarks. For example, Sun et al [19] are pioneers in adopting Deep Convolutional Neural Network (DCNN) in the regression framework to locate five fiducial facial landmarks.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Face Alignment via Multi-Regressors Collaborative Optimization

Liu¹,

Feng²,

Wang³

et al. 2019

IEEE Access

View full text Add to dashboard Cite

Face alignment is a fundamental step in facial image analysis. To solve the non-convex optimization problem, most cascade-based regression approaches conventionally utilize a single regressor to cover the entire optimization space. These measures are prone to average conflicting gradient directions, especially when applied to faces in the unconstrained condition with various poses and expressions. In this paper, we present an effective face alignment approach based on the multi-regressors collaborative optimization. The foundation of our method is the cascaded regression (CR) that has recently established itself as one of the most practical and effective frameworks for localizing the facial landmarks. CR is interpreted as a learning-based approach to iteratively optimize an objective function. On this basis, in all iterations, the proposed algorithm further divides the sample space into several clusters, in each of which, samples with similar gradient directions and one separate local regressor are learned. During the prediction stage, the unseen landmarks of a face image are evaluated by a linear combination of estimations from all cluster regressors with different weights. The experimental results demonstrate the advantages of our method on the general unconstrained images. INDEX TERMS Face alignment, multi-regressors, cascaded shape regression.

show abstract