# Scaling Up Computer Vision Neural Networks Using Fast Fourier Transform

## TL;DR

This paper explores the use of Fast Fourier Transform to efficiently scale up convolutional neural networks and Vision Transformers for high-resolution image processing, addressing computational challenges.

## Contribution

It introduces FFT-based methods to improve scalability of large kernels and high-resolution inputs in computer vision models.

## Key findings

- FFT accelerates large kernel convolutions
- FFT reduces complexity for high-resolution image processing
- Proposed methods improve model scalability

## Abstract

Deep Learning-based Computer Vision field has recently been trying to explore larger kernels for convolution to effectively scale up Convolutional Neural Networks. Simultaneously, new paradigm of models such as Vision Transformers find it difficult to scale up to larger higher resolution images due to their quadratic complexity in terms of input sequence. In this report, Fast Fourier Transform is utilised in various ways to provide some solutions to these issues.