DeCaFA: Deep Convolutional Cascade for Face Alignment in the Wild

Dapogny, Arnaud; Bailly, Kévin; Cord, Matthieu

doi:10.1109/iccv.2019.00699

Cited by 80 publications

(42 citation statements)

References 30 publications

(54 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…More recently, Deep Neural Networks and especially Convolutional Neural Networks approaches have gained attention and become popular tools in computer vision tasks [43], including face alignment. They show robust and accurate inferences of the facial landmarks [44][45][46][47][48][49][50]. CNNs can extract high-level image features, modeling complex nonlinear relationships between the facial appearance and the face shape.…”

Section: Discriminative Methodsmentioning

confidence: 99%

“…They are able to carry out several tasks at the same time like the pose estimation or a 3D shape deformable model, computing the 2D face shape as a projection of the 3D model [51,52]. Many of these approaches use heatmaps as a probability distribution map to compute accurately the coordinates of the facial landmark points in the input image [46,50,53,54].…”

Section: Discriminative Methodsmentioning

confidence: 99%

“…The current face alignment state-of-the-art has placed the automatic face landmarks detection close to the domain of the human perception in terms of accuracy [46,[48][49][50], especially for still images in nonchallenging conditions.…”

Section: Robust and Fast Face Alignment Systemmentioning

confidence: 99%

See 2 more Smart Citations

Real-time face alignment: evaluation methods, training strategies and implementation optimization

Casado

López

2021

J Real-Time Image Proc

View full text Add to dashboard Cite

Face alignment is a crucial component in most face analysis systems. It focuses on identifying the location of several keypoints of the human faces in images or videos. Although several methods and models are available to developers in popular computer vision libraries, they still struggle with challenges such as insufficient illumination, extreme head poses, or occlusions, especially when they are constrained by the needs of real-time applications. Throughout this article, we propose a set of training strategies and implementations based on data augmentation, software optimization techniques that help in improving a large variety of models belonging to several real-time algorithms for face alignment. We propose an extended set of evaluation metrics that allow novel evaluations to mitigate the typical problems found in real-time tracking contexts. The experimental results show that the generated models using our proposed techniques are faster, smaller, more accurate, more robust in specific challenging conditions and smoother in tracking systems. In addition, the training strategy shows to be applicable across different types of devices and algorithms, making them versatile in both academic and industrial uses.

show abstract

Section: Discriminative Methodsmentioning

confidence: 99%

Section: Discriminative Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Real-time face alignment: evaluation methods, training strategies and implementation optimization

Casado

López

2021

J Real-Time Image Proc

View full text Add to dashboard Cite

show abstract

“…DSRN [18] is a direct shape regression network for end-to-end face alignment by jointly handling the highly nonlinear relationship between face images and associated facial shapes in a unified framework. DeCaFA [19] is an end-to-end deep convolutional cascade architecture for face alignment; it uses fully-convolutional stages to keep full spatial resolution throughout the cascade and significantly outperforms existing approaches on challenging databases. Wang et al [20,21] put forward the idea of combining the face GAN network with the cascaded network to improve the face alignment algorithm, realize the accurate positioning of key points of the face, and solve the problem of facial expression lighting and occlusion for face detection.…”

Section: Related Workmentioning

confidence: 99%

Facial Landmark Detection Using Generative Adversarial Network Combined with Autoencoder for Occlusion

Liu

Zheng

et al. 2020

Mathematical Problems in Engineering

View full text Add to dashboard Cite

The performance of the facial landmark detection model will be in trouble when it is under occlusion condition. In this paper, we present an effective framework with the objective of addressing the occlusion problem for facial landmark detection, which includes a generative adversarial network with improved autoencoders (GAN-IAs) and deep regression networks. In this model, GAN-IA can restore the occluded face region by utilizing skip concatenation among feature maps to keep more details. Meanwhile, self-attention mechanism that is effective in modeling long-range dependencies is employed to recover harmonious images for occluded faces. Deep regression networks are used to learn a nonlinear mapping from facial appearance to facial shape. Benefited from the mutual cooperation of GAN-IA and deep regression networks, a robust facial landmark detection model is achieved for the occlusion problem and the performance of the model achieves obviously improvement on challenging datasets.

show abstract

“…There are also recent improvements in language understanding with Bidirectional Encoder Representations from Transformers (BERT) [9] and A Robustly Optimized BERT Pretraining Approach (RoBERTa) [10], and, adding to that, the recent breakthrough in task agnostic transfer learning by Howard et al [11]. In CV, DL has advanced, inter alia, the tasks of image classification [12,13], object-detection [14][15][16], object-tracking [17], pose estimation [18][19][20][21], superresolution [22], and semantic segmentation [23][24][25][26][27][28]. These advancements give rise to new applications in, e.g., solid-state materials science and chemical sciences [29,30], meteorology [31], medicine [32][33][34][35][36][37][38][39], seismology [40][41][42], biology [43], life sciences in general [44], chemistry [45], and physics [46][47][48][49][50][51][52]…”

Section: Introductionmentioning

confidence: 99%

Generalized Sparse Convolutional Neural Networks for Semantic Segmentation of Point Clouds Derived from Tri-Stereo Satellite Imagery

et al. 2020

View full text Add to dashboard Cite

We studied the applicability of point clouds derived from tri-stereo satellite imagery for semantic segmentation for generalized sparse convolutional neural networks by the example of an Austrian study area. We examined, in particular, if the distorted geometric information, in addition to color, influences the performance of segmenting clutter, roads, buildings, trees, and vehicles. In this regard, we trained a fully convolutional neural network that uses generalized sparse convolution one time solely on 3D geometric information (i.e., 3D point cloud derived by dense image matching), and twice on 3D geometric as well as color information. In the first experiment, we did not use class weights, whereas in the second we did. We compared the results with a fully convolutional neural network that was trained on a 2D orthophoto, and a decision tree that was once trained on hand-crafted 3D geometric features, and once trained on hand-crafted 3D geometric as well as color features. The decision tree using hand-crafted features has been successfully applied to aerial laser scanning data in the literature. Hence, we compared our main interest of study, a representation learning technique, with another representation learning technique, and a non-representation learning technique. Our study area is located in Waldviertel, a region in Lower Austria. The territory is a hilly region covered mainly by forests, agriculture, and grasslands. Our classes of interest are heavily unbalanced. However, we did not use any data augmentation techniques to counter overfitting. For our study area, we reported that geometric and color information only improves the performance of the Generalized Sparse Convolutional Neural Network (GSCNN) on the dominant class, which leads to a higher overall performance in our case. We also found that training the network with median class weighting partially reverts the effects of adding color. The network also started to learn the classes with lower occurrences. The fully convolutional neural network that was trained on the 2D orthophoto generally outperforms the other two with a kappa score of over 90% and an average per class accuracy of 61%. However, the decision tree trained on colors and hand-crafted geometric features has a 2% higher accuracy for roads.

show abstract

DeCaFA: Deep Convolutional Cascade for Face Alignment in the Wild

Cited by 80 publications

References 30 publications

Real-time face alignment: evaluation methods, training strategies and implementation optimization

Real-time face alignment: evaluation methods, training strategies and implementation optimization

Facial Landmark Detection Using Generative Adversarial Network Combined with Autoencoder for Occlusion

Generalized Sparse Convolutional Neural Networks for Semantic Segmentation of Point Clouds Derived from Tri-Stereo Satellite Imagery

Contact Info

Product

Resources

About