One millisecond face alignment with an ensemble of regression trees

By Vahid Kazemi, Josephine Sullivan. 2014

2,798 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

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

Citations in the corpus

A Style-Based Generator Architecture for Generative Adversarial Networks (2018, background): “heriting all the biases of that website) and automatically aligned [29] and cropped.”
Attribute2Image: Conditional Image Generation from Visual Attributes (2015, methodology): “Therefore, we perform frontalization [14, 17] (face regions are detected and normalized to the frontal view) on the faces returned by above-mentioned models.”
How Image Degradations Affect Deep CNN-Based Face Recognition? (2016, methodology): “Firstly, to model the sunglasses effect, the dark ellipse shape is fitted on the eye area by using landmark points, which are detected by dlib library [19] (see Fig.”
RotDCF: Decomposition of Convolutional Filters for Rotation-Equivariant Deep Networks (2018, methodology): “A CNN and a RotDCF Net (Table 7) are trained respectively using the gallery images from the 500 known subjects, which are preprocessed to be near-frontal and upright-only by aligning facial landmarks [18].”
VoxCeleb: A Large-Scale Speaker Identification Dataset (2017, methodology): “Facial landmark positions are detected for each face detection using the regression tree based method of [33].”
Age Progression/Regression by Conditional Adversarial Autoencoder (2017, methodology): “We use the face detection algorithm with 68 landmarks [11, 5] to crop out and align the faces, making the training more attainable.”
Deep Expectation of Real and Apparent Age from a Single Image Without Facial Landmarks (2016, methodology): “2014) and face landmark detection (Kazemi and Sullivan 2014) and six CNNs for classification in three age groups and for local (part face) and global (whole face) prediction of age.”
Face alignment by coarse-to-fine shape searching (2015, background): “Efficient learning-based feature has been proposed in [29] and [20].”
FSA-Net: Learning Fine-Grained Structure Aggregation for Head Pose Estimation From a Single Image (2019, methodology): “Most single-frame pose estimation methods utilize facial landmark detection for estimating head poses [20, 3].”
HiFaceGAN: Face Renovation via Collaborative Suppression and Replenishment (2020, methodology): “delity and perceptual realism of renovated faces. For semantic fidelity, we measure the feature embedding distance (FED) and landmark localization error (LLE) with a pretrained face recognition model [21], where the average L2 norm between feature embeddings is adopted for both metrics. For perceptu”
Geometry-Contrastive GAN for Facial Expression Transfer. (2018, methodology): “In our experiments, 68 facial landmark points are obtained through dlib 1 which implements the method proposed in [12].”
Unrestricted Facial Geometry Reconstruction Using Image-to-Image Translation (2017, background): “• Facial landmarks [25, 46, 32, 47] - a set of automatically detected key points on the face such as the tip of the nose and the corners of the eyes, which can guide the reconstruction process [49, 26, 1, 12, 29].”
On Face Segmentation, Face Swapping, and Face Perception (2017, methodology): “We use the method of [18] to detect faces and facial landmarks in each of the frames.”
Region Attention Networks for Pose and Occlusion Robust Facial Expression Recognition (2019, methodology): “In all the following experiments, we use the CNN detector and the ERT [29] based face alignment method in Dlib toolbox 1 to crop and align faces, and then resize them to the 1 http://dlib.net size of 224 × 224.”
Celeb-DF: A Large-Scale Challenging Dataset for DeepFake Forensics (2019, methodology): “The landmarks are used to align the faces to a standard conﬁguration [22].”
Learning Pose-Aware Models for Pose-Invariant Face Recognition in the Wild (2019, methodology): “These are: 1) DLIB [56] which provides 68 facial landmarks, 2) TDCNN [57] detects only ﬁve sparse landmarks, and 3) CLNF [47], [58] for 68 point detection.”
Positional Encoding as Spatial Inductive Bias in GANs (2020, background): “Table 2: Results based on 256(2) StyleGAN2 with a channel multiplier of two [22] in FFHQ dataset [21, 24].”
EmotioNet: An Accurate, Real-Time Algorithm for the Automatic Annotation of a Million Facial Expressions in the Wild (2016, methodology): “Specifically, we combine the algorithms defined in [24, 9] to automatically detect the 66 landmarks shown in Figure 2a.”