Covariance descriptor based convolution neural network for saliency computation in low contrast images

Xu, Xin; Mu, Nan; Zhang, Xiaolong; Li, Bo

doi:10.1109/ijcnn.2016.7727257

Cited by 8 publications

(1 citation statement)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Only very few works have considered using RCDs in conjunction with deep learning. In particular, [41] designed a CNN taking RCDs as input for the task of saliency computation. The focus of this work, however, differs fundamentally from ours, as it rather aims to process pre-computed RCDs, whereas we seek to learn second-order statistics from images.…”

Section: Cnns For Visual Recognitionmentioning

confidence: 99%

Second-order Convolutional Neural Networks

Yu¹,

Salzmann²

2017

Preprint

View full text Add to dashboard Cite

Convolutional Neural Networks (CNNs) have been successfully applied to many computer vision tasks, such as image classification. By performing linear combinations and element-wise nonlinear operations, these networks can be thought of as extracting solely first-order information from an input image. In the past, however, second-order statistics computed from handcrafted features, e.g., covariances, have proven highly effective in diverse recognition tasks. In this paper, we introduce a novel class of CNNs that exploit second-order statistics. To this end, we design a series of new layers that (i) extract a covariance matrix from convolutional activations, (ii) compute a parametric, second-order transformation of a matrix, and (iii) perform a parametric vectorization of a matrix. These operations can be assembled to form a Covariance Descriptor Unit (CDU), which replaces the fully-connected layers of standard CNNs. Our experiments demonstrate the benefits of our new architecture, which outperform the first-order CNNs, while relying on up to 90% fewer parameters.

show abstract

Section: Cnns For Visual Recognitionmentioning

confidence: 99%