HLearning / ai_papers

Data

CutOut

• Date: 2017/08
• Paper: Improved Regularization of Convolutional Neural Networks with Cutout

MixUP

• Date: 2017/10
• Paper: mixup: Beyond Empirical Risk Minimization

CutMix

• Date: 2019/05
• Paper: CutMix: Regularization Strategy to Train Strong Classifierswith Localizable Features

Base

GIoU

• Date: 2019/02
• Paper: Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression

DIoU

• Date: 2019/11
• Paper: Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression

Image Classification

VGG

• Date: 2014/09
• Paper: Very Deep Convolutional Networks for Large-Scale Image Recognition
• Code(keras):https://github.com/keras-team/keras-applications

ResNet

• Date: 2015/12
• Paper: Deep Residual Learning for Image Recognition
• Code(keras):https://github.com/keras-team/keras-applications

ResNetV2

• Date: 2016/03
• Paper: Identity Mappings in Deep Residual Networks
• Code(keras):https://github.com/keras-team/keras-applications

ResNeXt

• Date: 2016/11
• Paper: Aggregated Residual Transformations for Deep Neural Networks
• Code(keras):https://github.com/keras-team/keras-applications

InceptionV1

• Date: 2014/09
• Paper: Going Deeper with Convolutions

InceptionV2

• Date: 2015/02
• Paper: Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

InceptionV3

• Date: 2015/12
• Paper: Rethinking the Inception Architecture for Computer Vision
• Code(keras):https://github.com/keras-team/keras-applications

InceptionV4/InceptionResNetV2

• Date: 2016/02
• Paper: Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning
• Code(keras):https://github.com/keras-team/keras-applications

DenseNet

• Date: 2016/08
• Paper: Densely Connected Convolutional Networks
• Code(keras):https://github.com/keras-team/keras-applications

Xecption

• Date: 2016/10
• Paper: Xception: Deep Learning with Depthwise Separable Convolutions
• Code(keras):https://github.com/keras-team/keras-applications

MobileNet

• Date: 2017/04
• Paper: MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• Code(keras):https://github.com/keras-team/keras-applications

NASNet

• Date: 2017/07
• Paper: Learning Transferable Architectures for Scalable Image Recognition
• Code(keras):https://github.com/keras-team/keras-applications

EfficientNet

• Date: 2019/05
• Paper: EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
• Code(keras):https://github.com/keras-team/keras-applications

Object Detection

RCNN

• Date: 2013/11
• Paper: Rich feature hierarchies for accurate object detection and semantic segmentation

Fast RCNN

• Date: 2015/04
• Paper: Fast R-CNN

Faster RCNN

• Date: 2015/06
• Paper: Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

SSD

• Date: 2015/12
• Paper: SSD: Single Shot MultiBox Detector

YOLO_V1

• Date: 2015/06
• Paper: You Only Look Once: Unified, Real-Time Object Detection

YOLO_V2

• Date: 2016/12
• Paper: YOLO9000: Better, Faster, Stronger

YOLO_V3

• Date: 2018/04
• Paper: YOLOv3: An Incremental Improvement
• Code(darknet): https://github.com/pjreddie/darknet

YOLO_V4

• Date: 2020/04
• Paper: YOLOv4: Optimal Speed and Accuracy of Object Detection
• Code(darknet): https://github.com/AlexeyAB/darknet

RetinaNet

• Date: 2017/08
• Paper: Focal Loss for Dense Object Detection

CenterNet

• Date: 2019/04
• Paper: Objects as Points

MatrixNet

• Date: 2019/08
• Paper: Matrix Nets: A New Deep Architecture for Object Detection

Semantic Segmentation

UNet

• Date: 2015/05
• Paper: U-Net: Convolutional Networks for Biomedical Image Segmentation

SegNet

• Date: 2015/11
• Paper: SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation

FCN

• Date: 2014/12
• Paper: Fully Convolutional Networks for Semantic Segmentation

DeepLabV1

• Date: 2014/12
• Paper: Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs

DeepLabV2

• Date: 2016/06
• Paper: DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs

DeepLabV3

• Date: 2017/06
• Paper: Rethinking Atrous Convolution for Semantic Image Segmentation

DeepLabV3+

• Date: 2018/02
• Paper: Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation

Instance Segmentation

MaskRCNN

• Date: 2017/03
• Paper: Mask R-CNN

GAN

GAN

• Date: 2014/06
• Paper: Generative Adversarial Networks

DCGAN

• Date: 2015/11
• Paper: Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

InfoGAN

• Date: 2016/06
• Paper: InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets

CycleGAN

• Date: 2017/03
• Paper: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
• Code(pytorch): https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix

StyleGAN

• Date: 2018/12
• Paper: A Style-Based Generator Architecture for Generative Adversarial Networks
• Code(tensorflow): https://github.com/NVlabs/stylegan

SRGAN

• Date: 2016/09
• Paper: Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network

ESRGAN

• Date: 2019/09
• Paper: ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks