Deep Convolutional Network Cascade for Facial Point Detection

By Yi Sun, Xiaogang Wang, Xiaoou Tang. 2013

1,418 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Model Architectures

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 Model Architectures

Citations in the corpus

Deep Learning Face Attributes in the Wild (2014, methodology): “Thus the features are extracted at wrong positions [25].”
DeepPose: Human Pose Estimation via Deep Neural Networks (2013, neutral): “On the related problem of face pose estimation, Osadchy et al. [17] employ a NN-based pose embedding trained with a contrastive loss.”
Part Detector Discovery in Deep Convolutional Neural Networks (2014, neutral): “In [21] and [22], the part positions are estimated by a CNN speciﬁcally trained for this task.”
Not All Pixels Are Equal: Difficulty-Aware Semantic Segmentation via Deep Layer Cascade (2017, background): “The number of stages is three, which is a common setting in previous cascade methods [18, 31, 33].”
Deep learning for visual understanding (2017, positive): “[72] proposed three-level carefully designed convolutional networks to detect facial keypoints.”
3D Human Pose Estimation from Monocular Images with Deep Convolutional Neural Network (2014, background): “In [20], a cascade neural network is proposed for stage-by-stage prediction of facial points.”
DeepID-Net: Object Detection with Deformable Part Based Convolutional Neural Networks (2017, neutral): “Because of the power on learning feature representation, deep models have been widely used for object recognition, detection and other vision tasks [1], [6], [7], [19], [20], [21], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47],”
Robust Optimization for Deep Regression (2015, methodology): “…the regressed values correspond to a bounding box for localisation [16, 40], in human pose estimation, the values represent the positions of the body joints on the image plane [25, 31, 42], and in facial landmark detection, the predicted values denote the image locations of the facial points [39].”
Heterogeneous Multi-task Learning for Human Pose Estimation with Deep Convolutional Neural Network (2014, positive): “bor with the learned embedding features. In contrast to [23], we introduce accessory tasks for learning shared “pose features”, and output the joint locations directly from the regression network. In [22], a multi-stage system with deep convolutional networks is built for predicting facial point loc”
Dynamic Deep Neural Networks: Optimizing Accuracy-Efficiency Trade-offs by Selective Execution (2017, methodology): “For example, cascade methods [23, 29] rely on manually-selected thresholds to control execution; Dynamic Capacity Network [2] designs a way to directly calculate a saliency map for execution control.”
Multi-objective convolutional learning for face labeling (2015, methodology): “Given a test image, five facial keypoints on eyes, nose tip and two mouth corners are detected using the method in [22].”
Unsupervised Learning of Object Landmarks by Factorized Spatial Embeddings (2017, methodology): “Deep learning methods use cascaded CNNs [56], coarse-to-fine autoencoders [70], auxiliary attribute prediction [73, 74], learned deformations [69] and LSTMs [64].”
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].”
FSA-Net: Learning Fine-Grained Structure Aggregation for Head Pose Estimation From a Single Image (2019, result): “Deeplearning-based methods [48, 3, 38] estimate 3D face models using convolutional neural networks (CNNs) and gain superior performance compared to previous methods.”
Procedural Noise Adversarial Examples for Black-Box Attacks on Deep Convolutional Networks (2018, background): “Given their increased use in safety-critical and security applications such as autonomous vehicles [4, 74, 76], intrusion detection [28, 29], malicious string detection [65], and facial recognition [39, 70], it is important to ensure that such algorithms are robust to malicious adversaries.”
FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance (2023, neutral): “The concept of using multiple services for speed is known as model cascade Viola & Jones (2004), which has been applied in predictive ML domains such as pedestrian detection Cai et al. (2015) and facial recognition Li et al. (2015); Sun et al. (2013).”
Deep learning in robotics: a review of recent research (2017, background): “These applications include semantic image segmentation [52], object localization within images [53], scaling up images with super resolution [54], facial recognition [55, 56], scene recognition [57], and human gesture recognition [58].”