Translating or rotating an input image should not affect the results of many computervision tasks. Convolutional neural networks (CNNs) are already translation equivariant: input image translations produce proportion...
详细信息
ISBN:
(纸本)9781538604571
Translating or rotating an input image should not affect the results of many computervision tasks. Convolutional neural networks (CNNs) are already translation equivariant: input image translations produce proportionate feature map translations. this is not the case for rotations. Global rotation equivariance is typically sought through data augmentation, but patch-wise equivariance is more difficult. We present Harmonic Networks or H-Nets, a CNN exhibiting equivariance to patch-wise translation and 360-rotation. We achieve this by replacing regular CNN filters with circular harmonics, returning a maximal response and orientation for every receptive field patch. H-Nets use a rich, parameter-efficient and fixed computational complexity representation, and we show that deep feature maps within the network encode complicated rotational invariants. We demonstrate that our layers are general enough to be used in conjunction withthe latest architectures and techniques, such as deep supervision and batch normalization. We also achieve state-of-the-art classification on rotated-MNIST, and competitive results on other benchmark challenges.
Sparsity and low-rank structures are recently considered as an important property in various signal processing problems. they have been widely applied in image processing, communication, computervision, pattern recog...
详细信息
Sparsity and low-rank structures are recently considered as an important property in various signal processing problems. they have been widely applied in image processing, communication, computervision, patternrecognition, radar, etc. the main purpose of this paper is to provide a review on sparse representation and low-rank approximation, and their applications in sensor signal processing. three specific scenarios of sensor signal processing are further discussed. Simulations and experiments are presented in each signal processing scenario to demonstrate the capability of sparse representation and low-rank approximation.
暂无评论