LeViT: a Vision Transformer in ConvNet's Clothing for Faster Inference

04/02/2021
by   Ben Graham, et al.
0

We design a family of image classification architectures that optimize the trade-off between accuracy and efficiency in a high-speed regime. Our work exploits recent findings in attention-based architectures, which are competitive on highly parallel processing hardware. We re-evaluated principles from the extensive literature on convolutional neural networks to apply them to transformers, in particular activation maps with decreasing resolutions. We also introduce the attention bias, a new way to integrate positional information in vision transformers. As a result, we propose LeVIT: a hybrid neural network for fast inference image classification. We consider different measures of efficiency on different hardware platforms, so as to best reflect a wide range of application scenarios. Our extensive experiments empirically validate our technical choices and show they are suitable to most architectures. Overall, LeViT significantly outperforms existing convnets and vision transformers with respect to the speed/accuracy tradeoff. For example, at 80% ImageNet top-1 accuracy, LeViT is 3.3 times faster than EfficientNet on the CPU.

READ FULL TEXT

page 3

page 18

03/31/2021

Going deeper with Image Transformers

Transformers have been recently adapted for large scale image classifica...
05/21/2022

Vision Transformers in 2022: An Update on Tiny ImageNet

The recent advances in image transformers have shown impressive results ...
09/15/2022

On the Surprising Effectiveness of Transformers in Low-Labeled Video Recognition

Recently vision transformers have been shown to be competitive with conv...
09/16/2022

Quantum Vision Transformers

We design and analyse quantum transformers, extending the state-of-the-a...
07/05/2021

Vision Xformers: Efficient Attention for Image Classification

Although transformers have become the neural architectures of choice for...
11/27/2021

FQ-ViT: Fully Quantized Vision Transformer without Retraining

Network quantization significantly reduces model inference complexity an...
03/23/2018

SqueezeNext: Hardware-Aware Neural Network Design

One of the main barriers for deploying neural networks on embedded syste...