Benchmarking 3D Pose Estimation for Face Recognition

Dou, Pengfei; Wu, Yuhang; Shah, Shishir K.

doi:10.1109/icpr.2014.42

Cited by 9 publications

(3 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The literature on 3D face reconstruction is vast and the algorithm input ranges from depth map [KKT*14] and single image [TTHMM17,KZT*18,RSOEK17,DSK17,ZWC*20,CCZ*19, FWS*18,SSL*20,CLL*21,XYC*20,GSL*20] to multi‐view images [BCLT21, BCR*20, WBC*19,DYX*19, SBFB19, CLC*22] and videos [GZC*16,TBG*19]. Since our main focus is 3D face reconstruction from sparse‐view images using neural SDFs as the geometric representation, in this section, we briefly review 3D morphable models, multi‐view 3D face reconstruction methods, and the most relevant implicit neural representations.…”

Section: Related Workmentioning

confidence: 99%

Implicit Neural Deformation for Sparse‐View Face Reconstruction

Huang

Zheng

et al. 2022

Computer Graphics Forum

View full text Add to dashboard Cite

In this work, we present a new method for 3D face reconstruction from sparse‐view RGB images. Unlike previous methods which are built upon 3D morphable models (3DMMs) with limited details, we leverage an implicit representation to encode rich geometric features. Our overall pipeline consists of two major components, including a geometry network, which learns a deformable neural signed distance function (SDF) as the 3D face representation, and a rendering network, which learns to render on‐surface points of the neural SDF to match the input images via self‐supervised optimization. To handle in‐the‐wild sparse‐view input of the same target with different expressions at test time, we propose residual latent code to effectively expand the shape space of the learned implicit face representation as well as a novel view‐switch loss to enforce consistency among different views. Our experimental results on several benchmark datasets demonstrate that our approach outperforms alternative baselines and achieves superior face reconstruction results compared to state‐of‐the‐art methods.

show abstract

Section: Related Workmentioning

confidence: 99%

Implicit Neural Deformation for Sparse‐View Face Reconstruction

Huang

Zheng

et al. 2022

Computer Graphics Forum

View full text Add to dashboard Cite

show abstract

“…In the first category of methods, 2D coordinates of FPs (denotated as x a ) are first estimated using an automatic detector [3,35,21,2,28,37,39], then a projection matrix P is estimated to minimize a reprojection error between x a and the projected 2D locations computed by P•y a , where y a is a vector containing the coordinates of corresponding 3D fiducial points on Y 3D . Weakprospective projection model, golden standard approaches [10], POSIT [5], and other approaches [6] can be used to estimate the P based on x a and y a . Because this category of methods assumes the facial landmarks are detected, their performance relies heavily on automatic landmark detectors.…”

Section: Related Workmentioning

confidence: 99%

Facial 3D Model Registration Under Occlusions With SensiblePoints-based Reinforced Hypothesis Refinement

Wu,

Kakadiaris

2017

Preprint

Self Cite

View full text Add to dashboard Cite

Registering a 3D facial model to a 2D image under occlusion is difficult. First, not all of the detected facial landmarks are accurate under occlusions. Second, the number of reliable landmarks may not be enough to constrain the problem. We propose a method to synthesize additional points (SensiblePoints) to create pose hypotheses. The visual clues extracted from the fiducial points, non-fiducial points, and facial contour are jointly employed to verify the hypotheses. We define a reward function to measure whether the projected dense 3D model is well-aligned with the confidence maps generated by two fully convolutional networks, and use the function to train recurrent policy networks to move the SensiblePoints. The same reward function is employed in testing to select the best hypothesis from a candidate pool of hypotheses. Experimentation demonstrates that the proposed approach is very promising in solving the facial model registration problem under occlusion.

show abstract

“…Face alignment is essential for many facial analysis tasks such as facial expression detection [11,28], attribute recognition [9], face image processing [25], face recognition and verification [22,5,21,10,12]. Significant improvements have been achieved in face alignment with the emergence of discriminative regression methods [24,17,13].…”

Section: Introductionmentioning

confidence: 99%

Towards Arbitrary-View Face Alignment by Recommendation Trees

Zhu¹,

Cheng²,

Loy³

et al. 2015

Preprint

View full text Add to dashboard Cite

Learning to simultaneously handle face alignment of arbitrary views, e.g. frontal and profile views, appears to be more challenging than we thought. The difficulties lay in i) accommodating the complex appearance-shape relations exhibited in different views, and ii) encompassing the varying landmark point sets due to self-occlusion and different landmark protocols. Most existing studies approach this problem via training multiple viewpoint-specific models, and conduct head pose estimation for model selection. This solution is intuitive but the performance is highly susceptible to inaccurate head pose estimation. In this study, we address this shortcoming through learning an Ensemble of Model Recommendation Trees (EMRT), which is capable of selecting optimal model configuration without prior head pose estimation. The unified framework seamlessly handles different viewpoints and landmark protocols, and it is trained by optimising directly on landmark locations, thus yielding superior results on arbitrary-view face alignment. This is the first study that performs face alignment on the full AFLW dataset with faces of different views including profile view. State-of-the-art performances are also reported on MultiPIE and AFW datasets containing both frontal-and profile-view faces.

show abstract

Benchmarking 3D Pose Estimation for Face Recognition

Cited by 9 publications

References 19 publications

Implicit Neural Deformation for Sparse‐View Face Reconstruction

Implicit Neural Deformation for Sparse‐View Face Reconstruction

Facial 3D Model Registration Under Occlusions With SensiblePoints-based Reinforced Hypothesis Refinement

Towards Arbitrary-View Face Alignment by Recommendation Trees

Contact Info

Product

Resources

About