Axial-DeepLab: Stand-Alone Axial-Attention for Panoptic Segmentation

TL;DR

Axial-DeepLab introduces a novel axial-attention mechanism that reduces computational complexity and enhances global context modeling, achieving state-of-the-art results in panoptic segmentation and image classification.

Contribution

The paper proposes a position-sensitive axial-attention layer that enables fully attentional networks with reduced complexity, improving performance on multiple large-scale datasets.

Findings

Outperforms existing self-attention models on ImageNet

Achieves 2.8% PQ improvement on COCO test-dev

Offers a parameter- and computation-efficient variant

Abstract

Convolution exploits locality for efficiency at a cost of missing long range context. Self-attention has been adopted to augment CNNs with non-local interactions. Recent works prove it possible to stack self-attention layers to obtain a fully attentional network by restricting the attention to a local region. In this paper, we attempt to remove this constraint by factorizing 2D self-attention into two 1D self-attentions. This reduces computation complexity and allows performing attention within a larger or even global region. In companion, we also propose a position-sensitive self-attention design. Combining both yields our position-sensitive axial-attention layer, a novel building block that one could stack to form axial-attention models for image classification and dense prediction. We demonstrate the effectiveness of our model on four large-scale datasets. In particular, our model outperforms all existing stand-alone self-attention models on ImageNet. Our Axial-DeepLab improves 2.8% PQ over bottom-up state-of-the-art on COCO test-dev. This previous state-of-the-art is attained by our small variant that is 3.8x parameter-efficient and 27x computation-efficient. Axial-DeepLab also achieves state-of-the-art results on Mapillary Vistas and Cityscapes.