Dilated Convolution with Learnable Spacings: beyond bilinear interpolation

06/01/2023

∙

Dilated Convolution with Learnable Spacings (DCLS) is a recently proposed variation of the dilated convolution in which the spacings between the non-zero elements in the kernel, or equivalently their positions, are learnable. Non-integer positions are handled via interpolation. Thanks to this trick, positions have well-defined gradients. The original DCLS used bilinear interpolation, and thus only considered the four nearest pixels. Yet here we show that longer range interpolations, and in particular a Gaussian interpolation, allow improving performance on ImageNet1k classification on two state-of-the-art convolutional architectures (ConvNeXt and ConvFormer), without increasing the number of parameters. The method code is based on PyTorch and is available at https://github.com/K-H-Ismail/Dilated-Convolution-with-Learnable-Spacings-PyTorch

READ FULL TEXT

Dilated Convolution with Learnable Spacings: beyond bilinear interpolation

Dilated convolution with learnable spacings

Segmentation of Liver Lesions with Reduced Complexity Deep Models

Shape Adaptor: A Learnable Resizing Module

Learnable Heterogeneous Convolution: Learning both topology and strength

Entropy Transformer Networks: A Learning Approach via Tangent Bundle Data Manifold

Interpolating Points on a Non-Uniform Grid using a Mixture of Gaussians

Deep Feature Learning for Medical Acoustics

Dilated Convolution with Learnable Spacings: beyond bilinear interpolation

Related Research

Dilated convolution with learnable spacings

Segmentation of Liver Lesions with Reduced Complexity Deep Models

Shape Adaptor: A Learnable Resizing Module

Learnable Heterogeneous Convolution: Learning both topology and strength

Entropy Transformer Networks: A Learning Approach via Tangent Bundle Data Manifold

Interpolating Points on a Non-Uniform Grid using a Mixture of Gaussians

Deep Feature Learning for Medical Acoustics