GridMask
GridMask copied to clipboard
GridMask Data Augmentation
Introduction
This project is the code for implementing the GridMask augmentation for image classification and object detection. The full paper is available at: https://arxiv.org/abs/2001.04086.
Main Result
ImageNet
Baseline | + GridMask | |
---|---|---|
ResNet-50 | 76.5 | 77.9 |
ResNet-101 | 78.0 | 79.1 |
ResNet-152 | 78.3 | 79.7 |
COCO2017
Baseline | + GridMask | |
---|---|---|
FasterRCNN-R50-FPN | 37.4 | 39.2 |
FasterRCNN-X101-FPN | 41.2 | 42.6 |
You can find pretrained models to achieve the results in models