This is a PyTorch implementation of MogaNet from our paper:
Efficient Multi-order Gated Aggregation Network
Siyuan Li*, Zedong Wang*, Zicheng Liu, Chen Tan, Haitao Lin, Di Wu, Zhiyuan Chen, Jiangbin Zheng, and Stan Z. Li†. In arXiv, 2022.
We propose MogaNet, a new family of efficient ConvNets, to pursue informative context mining with preferable complexity-performance trade-offs.
Table of Contents
We plan to release implementations of MogaNet in a few months. Please watch us for the latest release. Currently, this repo is reimplemented according to our official implementations in OpenMixup, and we are working on cleaning up experimental results and code implementations.
- ImageNet-1K Training Code
- ImageNet-1K Fine-tuning Code
- Downstream Transfer to Object Detection and Instance Segmentation on COCO
- Downstream Transfer to Semantic Segmentation on ADE20K
- Image Classification on Google Colab and Web Demo
Please check INSTALL.md for installation instructions.
See TRAINING.md for ImageNet-1K training instructions, or refer to our OpenMixup implementations. We have released pre-trained models on OpenMixup in moganet-in1k-weights.
Model | resolution | Params(M) | Flops(G) | Top-1 (%) | Config | Download |
---|---|---|---|---|---|---|
MogaNet-XT | 224x224 | 2.97 | 0.80 | 76.5 | config / script | model / log |
MogaNet-T | 224x224 | 5.20 | 1.10 | 79.0 | config / script | model / log |
MogaNet-T | 256x256 | 5.20 | 1.44 | 79.6 | config / script | model / log |
MogaNet-T* | 256x256 | 5.20 | 1.44 | 80.0 | config / script | model / log |
MogaNet-S | 224x224 | 25.3 | 4.97 | 83.4 | config / script | model / log |
MogaNet-B | 224x224 | 43.9 | 9.93 | 84.2 | config / script | model / log |
MogaNet-L | 224x224 | 82.5 | 15.9 | 84.6 | config / script | model / log |
This project is released under the Apache 2.0 license.
This repository is built using the timm library, DeiT and ConvNeXt repositories.
If you find this repository helpful, please consider citing:
@article{Li2022MogaNet,
title={Efficient Multi-order Gated Aggregation Network},
author={Siyuan Li and Zedong Wang and Zicheng Liu and Cheng Tan and Haitao Lin and Di Wu and Zhiyuan Chen and Jiangbin Zheng and Stan Z. Li},
journal={ArXiv},
year={2022},
volume={abs/2211.03295}
}