Dilated Convolution with Learnable Spacings: beyond bilinear interpolation - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

Dilated Convolution with Learnable Spacings: beyond bilinear interpolation

Ismail Khalfaoui-Hassani
  • Fonction : Auteur
Thomas Pellegrini

Résumé

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

Dates et versions

hal-04236736 , version 1 (11-10-2023)

Identifiants

Citer

Ismail Khalfaoui-Hassani, Thomas Pellegrini, Timothée Masquelier. Dilated Convolution with Learnable Spacings: beyond bilinear interpolation. ICML Workshop: Differentiable Almost Everything, 2023, Hawaii, United States. ⟨hal-04236736⟩
22 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Mastodon Facebook X LinkedIn More