Please enable JavaScript.
Coggle requires JavaScript to display documents.
Data Augmentation, Augmentation Library, RandCrop, Manully designed…
Data Augmentation
Traditional Standard Methods
Image Mirroring
RandomHorizontalFlip, RandomVerticalFlip, RandomRotation, Transpose...
Color Transform
ColoeJitter, Grayscale
cropping and zooming
Resize, CenterCrop, RandomResizedCrop, TenCrop, FiveCrop...
Liner Transform
LinearTransform, RandomRotation, RandomPerspective
NAS based Image Transform
AutoAugment, Fast AutoArgument and Faster AutoAugment
paper:
https://arxiv.org/abs/1805.09501v1
. Code:
https://github.com/DeepVoltaire/AutoAugment
RandAugment
paper:
https://arxiv.org/pdf/1909.13719.pdf
. Code:
https://github.com/heartInsert/randaugment
GreedyAugment:
https://arxiv.org/abs/1908.00704
PBA
Image Cropping
CutOut
Paper:
https://arxiv.org/abs/1708.04552
. Code:
https://github.com/uoguelph-mlrg/Cutout
RandErasing
Paper:
https://arxiv.org/pdf/1708.04896.pdf
. Code:
https://github.com/zhunzhong07/Random-Erasing
HideAndSeek
Paper:
https://arxiv.org/pdf/1811.02545.pdf
. Code:
https://github.com/kkanshul/Hide-and-Seek
GridMask
Paper:
https://arxiv.org/abs/2001.04086
. Code:
https://github.com/akuxcw/GridMask
Batch Layer Image Aliasing
Mixup
Paper:
https://arxiv.org/abs/2001.04086
. Code:
https://github.com/akuxcw/GridMask
Cutmix
Paper:
https://arxiv.org/pdf/1905.04899v2.pdf
. Code:
https://github.com/clovaai/CutMix-PyTorch
Fmix
Augmentation
Library
albumentations：该库包含大量的传统图像数据增广方法，链接：
https://github.com/albumentations-team/albumentations
UDA：无监督数据增广，链接：
https://github.com/google-research/uda
torchsample：基于pytorch的高层封装，含数据增广模型训练等，链接：
https://github.com/ncullen93/torchsample
image_augmentor: 又一个传统图像数据增广方法，链接：
https://github.com/codebox/image_augmentor
imgaug：一个适合分类、检测的数据增广库。
https://github.com/aleju/imgaug
1.vidaug：视频数据增广方法，链接：
https://github.com/okankop/vidaug
pytorch-官方数据增广：
https://github.com/pytorch/vision/tree/master/torchvision/transforms
张航博士开源的FastAutoAugment：
https://github.com/zhanghang1989/Fast-AutoAug-Torch
FastAutoAugment官方实现：
https://github.com/kakaobrain/fast-autoaugment
AutoAugment的官方实现：
https://github.com/tensorflow/models/tree/master/research/autoaugment
Paddle官方数据增广：
https://github.com/PaddlePaddle/PaddleClas/tree/master/ppcls/data/imaug
RandCrop
Resize
RandFlip
Normalize
Transpose
Batch
Mixup/Cutmix
Cutout
NAS based Transform
Data augmentation based on search technology
Manully designed features