English

Spatially-Adaptive Filter Units for Compact and Efficient Deep Neural Networks

Computer Vision and Pattern Recognition 2020-02-07 v2

Abstract

Convolutional neural networks excel in a number of computer vision tasks. One of their most crucial architectural elements is the effective receptive field size, that has to be manually set to accommodate a specific task. Standard solutions involve large kernels, down/up-sampling and dilated convolutions. These require testing a variety of dilation and down/up-sampling factors and result in non-compact representations and excessive number of parameters. We address this issue by proposing a new convolution filter composed of displaced aggregation units (DAU). DAUs learn spatial displacements and adapt the receptive field sizes of individual convolution filters to a given problem, thus eliminating the need for hand-crafted modifications. DAUs provide a seamless substitution of convolutional filters in existing state-of-the-art architectures, which we demonstrate on AlexNet, ResNet50, ResNet101, DeepLab and SRN-DeblurNet. The benefits of this design are demonstrated on a variety of computer vision tasks and datasets, such as image classification (ILSVRC 2012), semantic segmentation (PASCAL VOC 2011, Cityscape) and blind image de-blurring (GOPRO). Results show that DAUs efficiently allocate parameters resulting in up to four times more compact networks at similar or better performance.

Keywords

Cite

@article{arxiv.1902.07474,
  title  = {Spatially-Adaptive Filter Units for Compact and Efficient Deep Neural Networks},
  author = {Domen Tabernik and Matej Kristan and Aleš Leonardis},
  journal= {arXiv preprint arXiv:1902.07474},
  year   = {2020}
}

Comments

Accepted for publication in International Journal of Computer Vision, Jan 02 2020

R2 v1 2026-06-23T07:45:49.731Z