Multi-objective convolutional learning for face labeling

By Sifei Liu, Jimei Yang, Chang Huang, Ming-Hsuan Yang. 2015

139 citations recorded by the source corpus.

This paper formulates face labeling as a conditional random field with unary and pairwise classifiers. We develop a novel multi-objective learning method that optimizes a single unified deep convolutional network with two distinct non-structured loss functions: one encoding the unary label likelihoods and the other encoding the pairwise label dependencies. Moreover, we regularize the network by using a nonparametric prior as new input channels in addition to the RGB image, and show that significant performance improvements can be achieved with a much smaller network size. Experiments on both the LFW and Helen datasets demonstrate state-of-the-art results of the proposed algorithm, and accurate labeling results on challenging images can be obtained by the proposed algorithm for real-world applications.

Subfields: Computer Vision and Pattern Recognition, Signal Processing

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

Convolutional Neural Fabrics (2016, background): “Moreover, [10, 19, 31] also include CRF and/or RBM models to encode spatial shape priors.”
Learning Affinity via Spatial Propagation Networks (2017, methodology): “Although many [27, 33, 20] achieve preferable performance, their results cannot be directly adopted by high-quality facial image editing applications.”
Fast Face-Swap Using Convolutional Neural Networks (2016, methodology): “This could be automated by combining our network with a neural network trained for segmentation [17, 25].”
Object Contour Detection with a Fully Convolutional Encoder-Decoder Network (2016, background): “Given the success of deep convolutional networks [31] for learning rich feature hierarchies, image labeling has been greatly advanced, especially on the task of semantic segmentation [11, 36, 34, 52, 40, 35].”
Learning Time/Memory-Efficient Deep Architectures with Budgeted Super Networks (2017, unknown): “op cost. (b1) Best CIFAR-10 architecture. (b2) Best Part Labels architecture (the output layer is the top-right layer in that case) Model Flop gain Error CNF 0 % 4.94 % CNF [18] 0 % 4.61 % Liu et. al [13] 0 % 4.76 % BSN 0.0 % 4.94 % 20.0 % 4.79 % 40.0 % 4.57 % >40 % 30.43 % Smart P-CNF 20.0 % 5.5”
Generative Face Completion (2017, methodology): “In addition, we also propose a face parsing network [14, 22, 13] as an additional loss to regularize the generation procedure and enforce a more reasonable and consistent result with contexts.”
From Facial Expression Recognition to Interpersonal Relation Prediction (2016, background): “tion [84,85,42,82,55], face parsing [50,47], face landmark detection [92,99,73], face attribute recognition [48,78,26], face recognition [66,58,68], and face clustering [94].”
On Face Segmentation, Face Swapping, and Face Perception (2017, methodology): “In [29] a network was trained to simultaneously segment multiple facial regions, including the entire face.”