Saved in:
Bibliographic Details
Main Authors: Chen, Chengpeng, Guo, Zichao, Zeng, Haien, Xiong, Pengfei, Dong, Jian
Format: Preprint
Published: 2022
Subjects:
Online Access:https://arxiv.org/abs/2211.06088
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913452941901824
author Chen, Chengpeng
Guo, Zichao
Zeng, Haien
Xiong, Pengfei
Dong, Jian
author_facet Chen, Chengpeng
Guo, Zichao
Zeng, Haien
Xiong, Pengfei
Dong, Jian
contents Feature reuse has been a key technique in light-weight convolutional neural networks (CNNs) architecture design. Current methods usually utilize a concatenation operator to keep large channel numbers cheaply (thus large network capacity) by reusing feature maps from other layers. Although concatenation is parameters- and FLOPs-free, its computational cost on hardware devices is non-negligible. To address this, this paper provides a new perspective to realize feature reuse implicitly and more efficiently instead of concatenation. A novel hardware-efficient RepGhost module is proposed for implicit feature reuse via reparameterization, instead of using concatenation operator. Based on the RepGhost module, we develop our efficient RepGhost bottleneck and RepGhostNet. Experiments on ImageNet and COCO benchmarks demonstrate that our RepGhostNet is much more effective and efficient than GhostNet and MobileNetV3 on mobile devices. Specially, our RepGhostNet surpasses GhostNet 0.5x by 2.5% Top-1 accuracy on ImageNet dataset with less parameters and comparable latency on an ARM-based mobile device. Code and model weights are available at https://github.com/ChengpengChen/RepGhost.
format Preprint
id arxiv_https___arxiv_org_abs_2211_06088
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle RepGhost: A Hardware-Efficient Ghost Module via Re-parameterization
Chen, Chengpeng
Guo, Zichao
Zeng, Haien
Xiong, Pengfei
Dong, Jian
Computer Vision and Pattern Recognition
Feature reuse has been a key technique in light-weight convolutional neural networks (CNNs) architecture design. Current methods usually utilize a concatenation operator to keep large channel numbers cheaply (thus large network capacity) by reusing feature maps from other layers. Although concatenation is parameters- and FLOPs-free, its computational cost on hardware devices is non-negligible. To address this, this paper provides a new perspective to realize feature reuse implicitly and more efficiently instead of concatenation. A novel hardware-efficient RepGhost module is proposed for implicit feature reuse via reparameterization, instead of using concatenation operator. Based on the RepGhost module, we develop our efficient RepGhost bottleneck and RepGhostNet. Experiments on ImageNet and COCO benchmarks demonstrate that our RepGhostNet is much more effective and efficient than GhostNet and MobileNetV3 on mobile devices. Specially, our RepGhostNet surpasses GhostNet 0.5x by 2.5% Top-1 accuracy on ImageNet dataset with less parameters and comparable latency on an ARM-based mobile device. Code and model weights are available at https://github.com/ChengpengChen/RepGhost.
title RepGhost: A Hardware-Efficient Ghost Module via Re-parameterization
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2211.06088