Face alignment through subspace constrained mean-shifts

By Jason M. Saragih, Simon Lucey, Jeffrey F. Cohn. IEEE International Conference on Computer Vision, 2009

415 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/5b65b449975d1444b2e0cbba6e780e2f71fc9f6e

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

Emotions Don't Lie: An Audio-Visual Deepfake Detection Method using Affective Cues (2020, background): “There is vast literature in inference of perceived emotions from a single modality or a combination of multiple modalities like facial expressions [10, 53], speech/audio signals [54], body pose [41], walking styles [16] and physiological features [32].”
Non-contact video-based pulse rate measurement on a mobile service robot (2014, methodology): “One of the approaches providing reliable and accurate fitting results under varying illumination is the Deformable Model Fitting by Regularized Landmark Mean-Shift (RLMS) approach introduced by Saragih et al. in [20].”
Recombinator Networks: Learning Coarse-to-Fine Feature Aggregation (2015, background): “The size of rectangle was drawn uniformly in the range [20, 50] pixels.”
Dog Breed Classification Using Part Localization (2012, background): “A good deal of work has also addressed the localization of ﬁducial points and parts of human faces [20–22 , 6].”
Deformable Model Fitting by Regularized Landmark Mean-Shift (2010, methodology): “we originally proposed in Saragih et al. (2009) .”
Localizing Parts of Faces Using a Consensus of Exemplars (2013, methodology): “While a number of approaches balance local feature detector responses on the image with prior global information about the feature configurations [6], [7], [13], [21], [25], [26], optimizing the resulting objective function remains a challenge.”
Large-Pose Face Alignment via CNN-Based Dense 3D Model Fitting (2016, background): “Recently, face alignment works are very popular in top vision venues, as demonstrated by the progress in Constrained Local Model based approaches [5, 22], AAM-based approaches [15, 13, 14] and regressionbased approaches [27, 4, 33].”
Fusing audio, visual and textual clues for sentiment analysis from multimodal content (2016, methodology): “GAVAM [56] was also used to extract facial expression features from the face.”
Supervised Descent Method and Its Applications to Face Alignment (2013, methodology): “[28] proposed a non-parametric representation to model the posterior likelihood and the resulting optimization method is reminiscent of mean-shift.”
Exemplar-Based Face Parsing (2013, methodology): “533 n/a n/a 0.425 0.472 0.455 0.687 n/a n/a Saragih et al. [18] 0.679 0.598 0.890 0.600 0.579 0.579 0.769 n/a 0.733 Liu et al. [12] 0.770 0.640 0.843 0.601 0.650 0.618 0.742 0.886 0.738 Gu & Kanade [4] 0.743 0.681 0.889 0.545 0.568 0.599 0.789 n/a 0.746 Ours, Steps 1 & 3 omitted 0.766 0.687 0.896 0.”
Coarse-to-Fine Auto-Encoder Networks (CFAN) for Real-Time Face Alignment (2014, methodology): “Generally speaking, the existing approaches can be categorized into holistic feature based methods [7,21,14,34,19,6] and local feature based methods [8,10,15,23,9,25,35,32,31,2,28,11].”
PIE (2020, methodology): “detection [Saragih et al. 2009] and facial recognition [King 2009] often fail”
A Main Directional Mean Optical Flow Feature for Spontaneous Micro-Expression Recognition (2016, methodology): “To build an effective feature for micro-expression recognition, we first detect and divide the face region into ROIs in video clips by utilizing a set of facial feature points obtained from an instance, called discriminative response map fitting (DRMF) [24], of the constrained local models [25], [26”