Awesome
octconv.pytorch
PyTorch implementation of Octave Convolution in Drop an Octave: Reducing Spatial Redundancy in Convolutional Neural Networks with Octave Convolution
ResNet-50/101 on ImageNet
| Architecture | LR decay strategy | Parameters | GFLOPs | Top-1 / Top-5 Accuracy (%) |
|---|---|---|---|---|
| ResNet-50 | step (90 epochs) | 25.557M | 4.089 | 76.010 / 92.834 |
| ResNet-50 | cosine (120 epochs) | 25.557M | 4.089 | 77.150 / 93.468 |
| Oct-ResNet-50 (alpha=0.5) | cosine (120 epochs) | 25.557M | 2.367 | 77.640 / 93.662 |
| ResNet-101 | cosine (120 epochs) | 44.549M | 7.801 | 78.898 / 94.304 |
| Oct-ResNet-101 (alpha=0.5) | cosine (120 epochs) | 44.549M | 3.991 | 78.794 / 94.330 |
| ResNet-152 | cosine (120 epochs) | 60.193M | 11.514 | 79.234 / 94.556 |
| Oct-ResNet-152 (alpha=0.5) | 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 | cosine (150 epochs) | 4.232M | 568.7M | 72.238 / 90.536 |
| Oct-MobileNetV1 | cosine (150 epochs) | 4.232M | 318.2M | 71.254 / 89.728 |
Acknowledgement
Official MXNet implmentation by @cypw
Citation
@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}
}