Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/d-li14/octconv.pytorch
PyTorch implementation of Octave Convolution with pre-trained Oct-ResNet and Oct-MobileNet models
https://github.com/d-li14/octconv.pytorch
deep-neural-networks iccv2019 imagenet mobilenet multi-scale octconv pytorch-implementation resnet
Last synced: 7 days ago
JSON representation
PyTorch implementation of Octave Convolution with pre-trained Oct-ResNet and Oct-MobileNet models
- Host: GitHub
- URL: https://github.com/d-li14/octconv.pytorch
- Owner: d-li14
- License: apache-2.0
- Created: 2019-04-27T11:10:53.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2020-07-08T00:45:02.000Z (over 4 years ago)
- Last Synced: 2024-12-08T10:42:02.454Z (17 days ago)
- Topics: deep-neural-networks, iccv2019, imagenet, mobilenet, multi-scale, octconv, pytorch-implementation, resnet
- Language: Python
- Homepage: https://arxiv.org/abs/1904.05049
- Size: 135 KB
- Stars: 291
- Watchers: 9
- Forks: 42
- Open Issues: 2
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# octconv.pytorch
[PyTorch](pytorch.org) implementation of Octave Convolution in [Drop an Octave: Reducing Spatial Redundancy in Convolutional Neural Networks with Octave Convolution](https://arxiv.org/abs/1904.05049)
## ResNet-50/101 on ImageNet
| Architecture | LR decay strategy | Parameters | GFLOPs | Top-1 / Top-5 Accuracy (%) |
| ------------------------ | ------------------- | ---------- | ------ | ----------------------- |
| [ResNet-50](https://drive.google.com/open?id=1n7H6WNrvtf0eyWeWotbWD1kb95iVWaze) | step (90 epochs) | 25.557M | 4.089 | 76.010 / 92.834 |
| [ResNet-50](https://drive.google.com/open?id=1_aconGn2oZB1Bvgq65g2tsqSI7CSPAEt) | cosine (120 epochs) | 25.557M | 4.089 | 77.150 / 93.468 |
| [Oct-ResNet-50 (alpha=0.5)](https://drive.google.com/open?id=1F9esqmbIJmfTOsAZ6_6JEUnI83LVgF_S) | cosine (120 epochs) | 25.557M | 2.367 | 77.640 / 93.662 |
| [ResNet-101](https://drive.google.com/file/d/128pkjPIN8hvjmbsSmUb62cikmiPjlgb1/view?usp=sharing) | cosine (120 epochs) | 44.549M | 7.801 | 78.898 / 94.304 |
| [Oct-ResNet-101 (alpha=0.5)](https://drive.google.com/file/d/1E3To8EZDlVX8EfIU4q4r6SHDUHStKOqG/view?usp=sharing) | cosine (120 epochs) | 44.549M | 3.991 | 78.794 / 94.330 |
| [ResNet-152](https://drive.google.com/file/d/1RZwAaVs1sUjUBlXTRR-zq0-S2d38Zb8Y/view?usp=sharing) | cosine (120 epochs) | 60.193M | 11.514 | 79.234 / 94.556 |
| [Oct-ResNet-152 (alpha=0.5)](https://drive.google.com/file/d/1lmNaN8W-ky91je2hbPfUv4va6Uj4_qw1/view?usp=sharing) | cosine (120 epochs) | 60.193M | 5.615 | 79.258 / 94.480 |
## MobileNet V1 on ImageNet
| Architecture | LR decay strategy | Parameters | FLOPs | Top-1 / Top-5 Accuracy (%) |
| ------------------------ | ------------------- | ---------- | ------ | ----------------------- |
| [MobileNetV1](https://drive.google.com/file/d/14FBekvITT77z2LX_2utFGMteK3gxN1vn/view?usp=sharing) | cosine (150 epochs) | 4.232M | 568.7M | 72.238 / 90.536 |
| [Oct-MobileNetV1](https://drive.google.com/file/d/1hpXYlHuTLeg04BOTFInrDYsaM4P-b0vK/view?usp=sharing) | cosine (150 epochs) | 4.232M | 318.2M | 71.254 / 89.728 |
## Acknowledgement
[Official MXNet implmentation](https://github.com/facebookresearch/OctConv) by [@cypw](https://github.com/cypw)
## Citation
```bibtex
@InProceedings{Chen_2019_ICCV,
author = {Chen, Yunpeng and Fan, Haoqi and Xu, Bing and Yan, Zhicheng and Kalantidis, Yannis and Rohrbach, Marcus and Yan, Shuicheng and Feng, Jiashi},
title = {Drop an Octave: Reducing Spatial Redundancy in Convolutional Neural Networks With Octave Convolution},
booktitle = {The IEEE International Conference on Computer Vision (ICCV)},
month = {October},
year = {2019}
}
```