Extensive Facial Landmark Localization with Coarse-to-Fine Convolutional Network Cascade

By Erjin Zhou, Haoqiang Fan, Zhimin Cao, Yuning Jiang, Qi Yin. 2013 IEEE International Conference on Computer Vision Workshops, 2013

368 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Signal Processing

View Paper: https://www.semanticscholar.org/paper/1a8ccc23ed73db64748e31c61c69fe23c48a2bb1

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 Signal Processing

Citations in the corpus

Mnemonic Descent Method: A Recurrent Process Applied for End-to-End Face Alignment (2016, methodology): “Our proposed method is also reminiscent of previously proposed deep learning methods for face alignment [43, 55, 47, 61, 63, 62].”
Unsupervised Learning of Object Landmarks by Factorized Spatial Embeddings (2017, methodology): “We follow [52] and use 3148 images from AFW [80], LFPW-train [3] and Helen-train [75] as training set, and 689 images from IBUG, LFPW-test and Helen-test as test set.”
Deep Alignment Network: A Convolutional Neural Network for Robust Face Alignment (2017, methodology): “While all of the above mentioned methods perform face alignment based only on local patches, there are some methods [35, 11] that estimate initial landmark positions using the entire face image and use local patches for refinement.”
Self-Supervised Multi-level Face Model Learning for Monocular Reconstruction at Over 250 Hz (2017, methodology): “, based on CNNs [58, 70, 15] or Restricted Boltzmann Machines [67].”
High-Resolution Representations for Labeling Pixels and Regions (2019, neutral): “Face++ [130] - 32.”
Self-supervised learning of a facial attribute embedding from video (2018, methodology): “For 300-W, 68 landmarks are regressed for 3148/689 train/test images which are obtained (as described in [51]) from combining multiple datasets [4, 67, 70].”
Large-Pose Face Alignment via CNN-Based Dense 3D Model Fitting (2016, methodology): “Figure 2 shows the overall process of the proposed method.”
Image and Vision Computing (2016, positive): “[48] (convolutional network framework), can continually achieve better results with continuous rise in the amount of training data.”
MoFA: Model-Based Deep Convolutional Face Autoencoder for Unsupervised Monocular Reconstruction (2017, methodology): “Various approaches are based on deep learning, including convolutional neural network (CNN) cascades [50, 63], a deep face shape model based on Restricted Boltzmann Machines [58], a recurrent network with longshort term memory [61], a recurrent encoder-decoder network for real-time facial landmark d”
Facial Landmark Detection: A Literature Survey (2018, methodology): “For example, in [122], similar cascaded CNN model is constructed to predict many more points (68 landmarks instead of 5).”
Face Alignment in Full Pose Range: A 3D Total Solution (2015, positive): “300 W [66] standardises multiple face alignment databases with 68 landmarks, including AFW [18], LFPW [67], HELEN [68], IBUG [66] and XM2VTS [69].”
Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network (2018, background): “In the beginning, face alignment that aims at detecting a special 2D fiducial points [1,2,3,4] is commonly used as a prerequisite for other facial tasks such as face recognition [5] and assists 3D face reconstruction [6,7] to a great extent.”
3D Face Reconstruction by Learning from Synthetic Data (2016, background): “As landmark detection algorithms have become faster and robust [13, 21, 25, 26] it is now common to use them for facial reconstruction.”
Pose2Seg: Detection Free Human Instance Segmentation (2018, background): “There has been much good research in the fields of human pose estimation [1, 2, 6, 14, 20, 26, 40], pedestrian detection [25, 41, 42], portrait segmentation [35, 36, 37], and face recognition [18, 23, 24, 27, 39, 43, 44], much of which has already produced practical value in real life.”
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].”
Bi-Real Net: Enhancing the Performance of 1-bit CNNs With Improved Representational Capability and Advanced Training Algorithm (2018, background): “tracking and alignment [24, 32, 34, 28, 27], object tracking [17], etc.”
Deciding How to Decide: Dynamic Routing in Artificial Neural Networks (2017, methodology): “This approach—dynamically routing signals through an inference system, based on their content—has already been incorporated into machine vision pipelines via methods such as boosting (Viola et al., 2005), coarse-to-ﬁne cascades (Zhou et al., 2013), and random decision forests (Ho, 1995).”
(2013) 300 Faces in-the-Wild Challenge: the first facial landmark localization challenge. In: 2013 IEEE International (2017, background): “[18] proposed a four-level convolutional network cascade, where each level was trained to locally refine the outputs of previous network levels.”