Look at Boundary: A Boundary-Aware Face Alignment Algorithm

By Wayne Wu, Chen Qian, Shuo Yang, Quan Wang, Yici Cai, Qiang Zhou. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018

511 citations recorded by the source corpus.

We present a novel boundary-aware face alignment algorithm by utilising boundary lines as the geometric structure of a human face to help facial landmark localisation. Unlike the conventional heatmap based method and regression based method, our approach derives face landmarks from boundary lines which remove the ambiguities in the landmark definition. Three questions are explored and answered by this work: 1. Why using boundary? 2. How to use boundary? 3. What is the relationship between boundary estimation and landmarks localisation? Our boundary-aware face alignment algorithm achieves 3.49% mean error on 300-W Fullset, which outperforms state-of-the-art methods by a large margin. Our method can also easily integrate information from other datasets. By utilising boundary information of 300-W dataset, our method achieves 3.92% mean error with 0.39% failure rate on COFW dataset, and 1.25% mean error on AFLW-Full dataset. Moreover, we propose a new dataset WFLW to unify training and testing across different factors, including poses, expressions, illuminations, makeups, occlusions, and blurriness. Dataset and model are publicly available at https://wywu.github.io/projects/LAB/LAB.html

Subfields: Computer Vision and Pattern Recognition, Dermatology

View Paper: https://www.semanticscholar.org/paper/e7265c560b3f10013bf70aacbbf0eb4631b7e2aa

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Dermatology

Citations in the corpus

LaFIn: Generative Landmark Guided Face Inpainting (2019, methodology): “Technically, G L can be accomplished by any landmark detector like [29, 14, 28].”
CelebV-HQ: A Large-Scale Video Facial Attributes Dataset (2022, methodology): “As stated before, we leverage [86] to detect the head pose in the yaw direction.”
General Facial Representation Learning in a Visual-Linguistic Manner (2021, methodology): “Comparing with state-of-the-art face alignment methods on WFLW test sets.”
Whole-Body Human Pose Estimation in the Wild (2020, background): “LAB [63] w/B 2.”
Adaptive Wing Loss for Robust Face Alignment via Heatmap Regression (2019, methodology): “For the WFLW dataset, we use the inter-ocular distance described in [62].”
AlphaPose: Whole-Body Regional Multi-Person Pose Estimation and Tracking in Real-Time (2022, methodology): “MPII [15] 16 ✓ ✓ 40K CrowdPose [19] 14 ✓ ✓ 80K PoseTrack [68] 15 ✓ ✓ 150K COCO [14] 17 ✓ ✓ 250K OneHand10K [70] 21 ✓ ✓ 10K FreiHand [62] 21 ✓ 130K MHP [71] 21 ✓ 80K WFLW [72] 98 ✓ ✓ 10K AFLW [73] 19 ✓ ✓ 25K COFW [74] 29 ✓ ✓ 1852 300W [61] 68 ✓ ✓ 3837 COCO-WholeBody [18] 133 ✓ ✓ ✓ ✓ 250K Halpe-FullBo”
Deep High-Resolution Representation Learning for Visual Recognition (2019, methodology): “Following [149], [191], we train our models on 20 , 000 training images, and report the results on the AFLW-Full set ( 4 , 386 testing images) and the AFLW-Frontal set ( 1314 testing images selected from 4386 testing images).”
3FabRec: Fast Few-Shot Face Alignment by Reconstruction (2019, methodology): “The recent advances in deep learning techniques [3] coupled with the availability of large, annotated databases have allowed steady progress with localization accuracy on a typical benchmark increasing by 100% (from [51] to [49] see below for more related work).”
DeeperForensics-1.0: A Large-Scale Dataset for Real-World Face Forgery Detection (2020, background): “We only use the facial regions (detected and cropped by LAB [40]) of the source data, so we can neglect the background.”
High-Resolution Representations for Labeling Pixels and Regions (2019, methodology): “The WFLW dataset [101] is a recently-built dataset based on the WIDER Face [112].”
Zoo-Tuning: Adaptive Transfer from a Zoo of Models (2021, methodology): “To explore the usage of Zoo-Tuning on more diverse and complex downstream vision tasks, we use the same model zoo as the image classification tasks in Section 4.2 and consider transferring to three facial landmark detection tasks, 300W (Sagonas et al., 2013), WFLW (Wu et al., 2018), and COFW (Burgos”
MEAD: A Large-Scale Audio-Visual Dataset for Emotional Talking-Face Generation (2020, methodology): “For the videos, we crop and align the face in each frame using facial landmarks detected with an open source face alignment tool [56].”