Dilated Convolution with Learnable Spacings: beyond bilinear interpolation

Abstract

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 Conv\-Former), 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

Published in ICML 2023 Workshop on Differentiable Almost Everything: Differentiable Relaxations, Algorithms, Operators, and Simulators. 2023

References in corpus (8)

• An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
• MetaFormer Baselines for Vision
• FlexConv: Continuous Kernel Convolutions with Differentiable Kernel Sizes
• Understanding Gaussian Attention Bias of Vision Transformers Using Effective Receptive Fields
• Dilated Convolution with Learnable Spacings
• SMPConv: Self-moving Point Representations for Continuous Convolution
• GMConv: Modulating Effective Receptive Fields for Convolutional Kernels
• Gaussian Mixture Convolution Networks