ReplaceBlock: An improved regularization method based on background information

TL;DR

ReplaceBlock is a novel regularization technique that enhances neural network training by simulating occlusion of target objects using background information, leading to improved performance over DropBlock.

Contribution

It introduces ReplaceBlock, a new regularization method that leverages background features to simulate occlusion, improving neural network robustness.

Findings

ReplaceBlock outperforms DropBlock in regularization effectiveness.

The method improves network robustness against occlusion.

Experimental results demonstrate better generalization with ReplaceBlock.

Abstract

Attention mechanism, being frequently used to train networks for better feature representations, can effectively disentangle the target object from irrelevant objects in the background. Given an arbitrary image, we find that the background's irrelevant objects are most likely to occlude/block the target object. We propose, based on this finding, a ReplaceBlock to simulate the situations when the target object is partially occluded by the objects that are deemed as background. Specifically, ReplaceBlock erases the target object in the image, and then generates a feature map with only irrelevant objects and background by the model. Finally, some regions in the background feature map are used to replace some regions of the target object in the original image feature map. In this way, ReplaceBlock can effectively simulate the feature map of the occluded image. The experimental results show that ReplaceBlock works better than DropBlock in regularizing convolutional networks.