Superpixel Mixing: A Data Augmentation Technique For Robust Deep Visual Recognition Models

Abstract

Data augmentation can mitigate overfitting problems in data exploration without increasing the size of the model. Existing

cutmix-based data augmentation has been proven to signifi- cantly enhance deep learning performance. However, many

existing methods overlook the discriminative local context of the image and rely on ad hoc regions consisting of square or rectangular local regions, resulting in the loss of complete

semantic object parts. In this work, we propose a superpixel- wise local-context-aware efficient image mixing approach

for data augmentation, aiming to overcome the limitations

previously mentioned. Our approach only requires one for- ward propagation using a superpixel attention-based label

mixing with lower computational complexity. The model is trained using a combination of a global classification of the mixed (augmented) image loss, a superpixel-wise weighted local classification loss, and a superpixel-based weighted contrastive learning loss. The last two losses are based on

the superpixel-aware attentive embeddings. Thus, the result- ing deep encoder can learn both local and global features of

the images, capturing object-part local context information. Experiments on diverse benchmarks, such as ImageNet-1K

and CUB-200-2011, indicate that the proposed method out- performs many augmentation methods for visual recognition.

We have not only demonstrated its effectiveness on CNN models, but also on transformer models.