https://github.com/scut-aitcm/Competitive-Inner-Imaging-SENet

Source code of paper: (not available now)
https://github.com/scut-aitcm/Competitive-Inner-Imaging-SENet

attention-mechanism computer-vision deep-learning mxnet residual-networks

Last synced: 2 days ago
JSON representation

Source code of paper: (not available now)

• Host: GitHub
• URL: https://github.com/scut-aitcm/Competitive-Inner-Imaging-SENet
• Owner: scut-aitcm
• Created: 2018-07-16T10:34:29.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-11-25T10:03:07.000Z (almost 6 years ago)
• Last Synced: 2024-05-22T04:18:14.387Z (6 months ago)
• Topics: attention-mechanism, computer-vision, deep-learning, mxnet, residual-networks
• Language: Python
• Homepage:
• Size: 3.71 MB
• Stars: 91
• Watchers: 14
• Forks: 11
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• Awesome-MXNet - **CompetitiveSENet**

README

        
# Competitive-Inner-Imaging-SENet

---

Source code of paper: 

   **(not availbale now)** 

---

## Architecture

|Competitive Squeeze-Exciation Architecutre for Residual block|

|-|

|![architecutre](pictures/architecture.png)|

---

SE-ResNet module and CMPE-SE-ResNet modules:

|Normal SE|Double FC squeezes|Conv 2x1 pair-view|Conv 1x1 pair-view|

|-|-|-|-|

|![](pictures/se_resnet_module.png)|![](pictures/cmpe_se_resnet_double_FC_squeeze.png)|![](pictures/cmpe_se_resnet_conv2x1.png)|![](pictures/cmpe_se_resnet_conv1x1.png)|

The Novel Inner-Imaging Mechanism for Channel Relation Modeling in Channel-wise Attention of ResNets (even All CNNs):

|Basic Inner-Imaing Mode|Folded Inner-Imaging Mode|

|-|-|

|![](pictures/Basic-Inner-Imaging.png)|![](pictures/Folded-Inner-Imaging.png)|

---

## Requirements

- **MXNet 1.2.0**

- Python 2.7

- CUDA 8.0+(for GPU)

---

## Citation

not available now

---

## Essential Results

Best record of this novel model on CIFAR-10 and CIFAR-100 (used "*mixup*" ([https://arxiv.org/abs/1710.09412](https://arxiv.org/abs/1710.09412))) can achieve: **97.55%** and **84.38%**.

 

The test result on Kaggle: [CIFAR-10 - Object Recognition in Images](https://www.kaggle.com/c/cifar-10) 

![](pictures/cifar10_kaggle.png)

Inner-Imaging Examples & Channel-wise Attention Outputs

![](pictures/appendix_a_fig1.png)