RepGhost: A Hardware-Efficient Ghost Module via Re-parameterization

TL;DR

This paper introduces RepGhost, a hardware-efficient module for CNNs that reuses features implicitly through reparameterization, improving efficiency and accuracy on mobile devices compared to existing models.

Contribution

The paper proposes a novel RepGhost module that enables implicit feature reuse via reparameterization, reducing hardware costs compared to concatenation-based methods.

Findings

RepGhostNet outperforms GhostNet 0.5x by 2.5% Top-1 accuracy on ImageNet.

RepGhostNet achieves similar latency with fewer parameters on mobile devices.

The method is more hardware-efficient than traditional concatenation-based feature reuse.

Abstract

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.