Random Erasing Data Augmentation

16 Aug 2017Zhun Zhong • Liang Zheng • Guoliang Kang • Shaozi Li • Yi Yang

In this paper, we introduce Random Erasing, a new data augmentation method for training the convolutional neural network (CNN). In training, Random Erasing randomly selects a rectangle region in an image and erases its pixels with random values. In this process, training images with various levels of occlusion are generated, which reduces the risk of over-fitting and makes the model robust to occlusion.

Full paper

Evaluation


Task Dataset Model Metric name Metric value Global rank Compare
Person Re-Identification DukeMTMC-reID SVDNet + Random Erasing Rank-1 79.3 # 3
Person Re-Identification DukeMTMC-reID SVDNet + Random Erasing MAP 62.4 # 3
Person Re-Identification DukeMTMC-reID TriNet + Random Erasing Rank-1 73.0 # 6
Person Re-Identification DukeMTMC-reID TriNet + Random Erasing MAP 56.6 # 6
Image Classification Fashion-MNIST Random Erasing Percentage error 3.65 # 1
Object Detection PASCAL VOC 2007 I+ORE MAP 76.2% # 11