https://github.com/afm-shahab-uddin/SaliencyMix

SaliencyMix: A Saliency Guided Data Augmentation Strategy for Better Regularization
https://github.com/afm-shahab-uddin/SaliencyMix

Last synced: about 2 months ago
JSON representation

SaliencyMix: A Saliency Guided Data Augmentation Strategy for Better Regularization

• Host: GitHub
• URL: https://github.com/afm-shahab-uddin/SaliencyMix
• Owner: afm-shahab-uddin
• Created: 2020-11-12T11:36:35.000Z (almost 4 years ago)
• Default Branch: main
• Last Pushed: 2022-11-23T02:56:05.000Z (almost 2 years ago)
• Last Synced: 2024-07-23T14:28:04.012Z (2 months ago)
• Language: Python
• Size: 67.4 KB
• Stars: 45
• Watchers: 2
• Forks: 13
• Open Issues: 2
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# SaliencyMix

SaliencyMix: A Saliency Guided Data Augmentation Strategy for Better Regularization

CIFAR training and testing code is based on 

- [Cutout](https://github.com/uoguelph-mlrg/Cutout)

The ImageNet is based on

- [Cutmix-PyTorch](https://github.com/clovaai/CutMix-PyTorch)

### Requirements  

- Python3

- PyTorch (> 1.0)

- torchvision (> 0.2)

- NumPy

- OpenCV-contrib-python (4.2.0.32)

### CIFAR

Please use "SaliencyMix_CIFAR" directory

#### CIFAR 10

-To train ResNet18 on CIFAR10 with SaliencyMix and traditional data augmentation:    

```

CUDA_VISIBLE_DEVICES=0,1 python saliencymix.py \

--dataset cifar10 \

--model resnet18 \

--beta 1.0 \

--salmix_prob 0.5 \

--batch_size 128 \

--data_augmentation \

--learning_rate 0.1

```

-To train ResNet50 on CIFAR10 with SaliencyMix and traditional data augmentation:    

```

CUDA_VISIBLE_DEVICES=0,1 python saliencymix.py \

--dataset cifar10 \

--model resnet50 \

--beta 1.0 \

--salmix_prob 0.5 \

--batch_size 128 \

--data_augmentation \

--learning_rate 0.1

```

-To train WideResNet on CIFAR10 with SaliencyMix and traditional data augmentation:    

```

CUDA_VISIBLE_DEVICES=0,1 python saliencymix.py \

--dataset cifar10 \

--model wideresnet \

--beta 1.0 \

--salmix_prob 0.5 \

--batch_size 128 \

--data_augmentation \

--learning_rate 0.1

```

#### CIFAR 100

-To train ResNet18 on CIFAR100 with SaliencyMix and traditional data augmentation:    

```

CUDA_VISIBLE_DEVICES=0,1 python saliencymix.py \

--dataset cifar100 \

--model resnet18 \

--beta 1.0 \

--salmix_prob 0.5 \

--batch_size 128 \

--data_augmentation \

--learning_rate 0.1

```

-To train ResNet50 on CIFAR100 with SaliencyMix and traditional data augmentation:    

```CUDA_VISIBLE_DEVICES=0,1 python saliencymix.py \

--dataset cifar100 \

--model resnet50 \

--beta 1.0 \

--salmix_prob 0.5 \

--batch_size 128 \

--data_augmentation \

--learning_rate 0.1

```

-To train WideResNet on CIFAR100 with SaliencyMix and traditional data augmentation:    

```

CUDA_VISIBLE_DEVICES=0,1 python saliencymix.py \

--dataset cifar100 \

--model wideresnet \

--beta 1.0 \

--salmix_prob 0.5 \

--batch_size 128 \

--data_augmentation \

--learning_rate 0.1

```

### ImageNet

-Please use "SaliencyMix-ImageNet" directory

### Train Examples

- ImageNet with 4 NVIDIA GeForce RTX 2080 Ti GPUs 

```

python train.py \

--net_type resnet \

--dataset imagenet \

--batch_size 256 \

--lr 0.1 \

--depth 50 \

--epochs 300 \

--expname ResNet50 \

-j 40 \

--beta 1.0 \

--salmix_prob 1.0 \

--no-verbose

```

### Test Examples using ImageNet Pretrained models

- Trained models can be downloaded from [here](https://drive.google.com/drive/folders/1vnJHtgzcBInuPZVkwQxQ5A5SE_i_-EON?usp=sharing)

- ResNet-50

```

python test.py \

--net_type resnet \

--dataset imagenet \

--batch_size 64 \

--depth 50 \

--pretrained /runs/ResNet50_SaliencyMix_21.26/model_best.pth.tar

```

- ResNet-101

```

python test.py \

--net_type resnet \

--dataset imagenet \

--batch_size 64 \

--depth 101 \

--pretrained /runs/ResNet101_SaliencyMix_20.09/model_best.pth.tar

```