Interlinked Convolutional Neural Networks for Face Parsing

By Yisu Zhou, Xiaolin Hu. 2015

68 citations recorded by the source corpus.

Face parsing is a basic task in face image analysis. It amounts to labeling each pixel with appropriate facial parts such as eyes and nose. In the paper, we present a interlinked convolutional neural network iCNN for solving this problem in an end-to-end fashion. It consists of multiple convolutional neural networks CNNs taking input in different scales. A special interlinking layer is designed to allow the CNNs to exchange information, enabling them to integrate local and contextual information efficiently. The hallmark of iCNN is the extensive use of downsampling and upsampling in the interlinking layers, while traditional CNNs usually uses downsampling only. A two-stage pipeline is proposed for face parsing and both stages use iCNN. The first stage localizes facial parts in the size-reduced image and the second stage labels the pixels in the identified facial parts in the original image. On a benchmark dataset we have obtained better results than the state-of-the-art methods.

Subfields: Computer Vision and Pattern Recognition

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

Convolutional Neural Fabrics (2016, methodology): “Finally, we analyze the number of response maps, parameters, and activations of fabrics in Section 3.3.”
Deep High-Resolution Representation Learning for Visual Recognition (2019, background): “Multi-scale fusion1 is widely studied [11], [19], [24], [42], [58], [66], [122], [123], [157], [161], [181], [188].”
Deep High-Resolution Representation Learning for Human Pose Estimation (2019, background): “There are related multi-scale networks for classification and segmentation [5, 8, 72, 78, 29, 73, 53, 54, 23, 80, 53, 51, 18].”
Semantic Segmentation using Adversarial Networks (2016, neutral): “Alternatively, one can use dilated convolutions to increase the receptive ﬁeld size without losing resolution [2, 34], skip connections to earlier high-resolution layers [16, 27], or multi-resolution networks [29, 36].”
Multi-Scale Dense Networks for Resource Efficient Image Classification (2017, background): “Our network architecture borrows elements from neural fabrics (Saxena & Verbeek, 2016) and others (Zhou et al., 2015; Jacobsen et al., 2017; Ke et al., 2016)”
High-Resolution Representations for Labeling Pixels and Regions (2019, background): “Representative works include GridNet [30], convolutional neural fabrics [86], interlinked CNNs [132], and the recently-developed high-resolution networks (HRNet) [91] that is our interest.”
Residual Conv-Deconv Grid Network for Semantic Segmentation (2017, methodology): “ 2 the conv-deconv network which does not have any residual connections. GridNet subsumes FRNN and can be seen as a generalisation of this network. The idea of networks with multiple paths is not new [8, 19, 27]. Zhou et al. studied a face parsing task with interlinked convolutional neural networks ”
HRFormer: High-Resolution Transformer for Dense Prediction (2021, methodology): “In the development of high-resolution convolutional neural networks, the community has developed three main paths including: (i) applying dilated convolutions to remove some down-sample layers [6, 51], (ii) recovering high-resolution representations from low-resolution representations with decoders ”