Active Appearance Models Revisited

By Iain Matthews, Simon Baker. International Journal of Computer Vision, 2004

1,802 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/978ae7f589f2b8aebc04655a0d9b19eabe97ce86

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

Appearance-based face recognition and light-fields (2004, methodology): “face model (a task we have been actively working on [18])”
Unsupervised Discovery of Object Landmarks as Structural Representations (2018, methodology): “For human faces, there are active appearance models [10, 38, 11], template-based methods [42, 83], regression-based methods [61, 12, 7, 48], and more recent methods based on deep neural networks [55, 77, 81, 82, 75, 70, 68, 33, 71].”
Reconstructing 3D Human Pose from 2D Image Landmarks (2012, methodology): “It has been successfully applied to constrained deformable objects, such as faces [4] and action-specific body reconstruction, such as walking, [5].”
Robust Face Landmark Estimation under Occlusion (2013, background): “In recent years, improvements over AAM and ASM have been proposed [31, 11, 32, 13, 37, 36].”
The First Facial Landmark Tracking in-the-Wild Challenge: Benchmark and Results (2015, background): “Rigid and non-rigid tracking of faces and facial features has been a very popular topic of research over the past twenty years [21, 29, 20, 17, 25, 37, 43, 23, 61, 19, 49, 14, 31, 34, 62, 63, 64, 10, 32, 40, 51, 39, 50, 35, 24, 26, 53, 56, 57, 54].”
2D vs. 3D Deformable Face Models: Representational Power, Construction, and Real-Time Fitting (2007, methodology): “Although computing the inverse compositional warp update is involved [18], it is not computationally demanding because it only requires operations on the mesh, rather than on the images.”
Learn to Combine Multiple Hypotheses for Accurate Face Alignment (2013, methodology): “The baseline is a project-out inverse compositional AAM method proposed in [16] with the edge-structure feature used in [5].”
Continuous sign language recognition: Towards large vocabulary statistical recognition systems handling multiple signers (2015, methodology): “[46] in 2004.”
Head Pose Estimation in Computer Vision: A Survey (2009, positive): “Fitting methods based on the inverse compositional image alignment algorithm overcome the linear assumption of how appearance error relates to the gradient descent search and allows for more accurate, real-time convergence [72].”
A deep learning approach for generalized speech animation (2017, methodology): “The video data of KB-2k is compactly parameterized using the coefficients of linear models of lower facial shape and appearance that an Active Appearance Model (AAM) optimizes to track the video frames [Cootes et al. 2001; Matthews and Baker 2004].”
Fine-Grained Head Pose Estimation Without Keypoints (2017, methodology): “torically, there have been several major approaches to face modeling, with two primary ones being discriminative/landmark-based approaches [25], [28] and parameterized appearance models, or PAMs [3], [14] (see [29] for additional discussion). In recent years, methods which directly extract 2D facial”
Adaptive Multimodal Fusion by Uncertainty Compensation With Application to Audiovisual Speech Recognition (2009, neutral): “We have shown that it is partly justified in the constant and modelaligned errors case; otherwise, one should use the more general rules in (16).”
State of the Art on Monocular 3D Face Reconstruction, Tracking, and Applications (2018, methodology): “Some methods are holistic, using the whole face region to localize the facial landmarks like Active Appearance Models (AAM) [CTCG95,CET01,MB04, XBMK04].”
Face alignment through subspace constrained mean-shifts (2009, positive): “This leads to better generalization with limited data compared to holistic representations [10, 11, 14, 15], since it needs only account for local correlations between pixel values.”
Photorealistic Facial Texture Inference Using Deep Neural Networks (2016, methodology): “[12] developed the widely used active appearance models (AAM) based on linear combinations of shape and appearance, which has resulted in several important subsequent works [1, 10, 33].”
Face alignment by coarse-to-fine shape searching (2015, methodology): “A number of methods have been proposed for face alignment, including the classic active appearance model [12, 22, 24] and constrained local model [13, 35, 31, 15].”
Look at Boundary: A Boundary-Aware Face Alignment Algorithm (2018, methodology): “In the literature of face alignment, besides classic meth-ods (ASMs [34, 23], AAMs [13, 41, 33, 25], CLMs [29, 42] and Cascaded Regression Models [7, 5, 58, 8, 72, 73, 18]), recently, state-of-the-art performance has been achieved with Deep Convolutional Neural Networks (DC-NNs).”
Detecting depression from facial actions and vocal prosody (2009, methodology): “Remaining frames were automatically aligned using a gradient-descent AAM fit described in [ 35 ]”