plugin style module for nerual network performance improvement

This repo amis at nerual network performance improvement with modification as few as possible. Plugin-style network module can almost insert into any networks and get expected improvement. Here's a collection list about the papers related to performance improvement, considering model parameters numbers / model size, model accuracy and inference speed.

Plugin network module

[CVPR 2018] [SENet] Squeeze-and-Excitation Networks paper code

[CVPR 2019] CSPNet: A New Backbone that can Enhance Learning Capability of CNN paper code

[CVPR 2018] Non-local Neural Networks paper code

Tricks for training

[CVPR 2019] Bag of Tricks for Image Classification with Convolutional Neural Networks paper code

[CVPR 2022] [ConNext] A ConvNet for the 2020s paper code

Light-weight network structure / Inference speed

ShuffleNet Series code

[CVPR 2018] ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices paper

[ECCV 2018] ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design paper

MobileNet Series code

[CVPR 2017] MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications paper

[CVPR 2018] MobileNetV2: Inverted Residuals and Linear Bottlenecks paper

[ICCV 2019] Searching for MobileNetV3 paper

MobileVit Series code

[ICLR 2022] MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer paper

note: code v2 already released, paper not yet

Others

[arxiv 2022] EdgeFormer: Improving Light-weight ConvNets by Learning from Vision Transformers paper code