BA-Net: Bridge Attention in Deep Neural Networks

TL;DR

This paper introduces bridge attention, a novel mechanism that enhances information flow between convolutional layers, leading to improved performance in image classification and other vision tasks.

Contribution

The paper proposes BAv2, an improved bridge attention model with an adaptive selection operator, boosting performance over existing attention methods in deep neural networks.

Findings

BAv2 achieves 80.49% Top-1 accuracy on ImageNet with ResNet50.

BAv2 surpasses baseline models by over 1.5% in accuracy.

Integration of BAv2 improves various vision tasks across different architectures.

Abstract

Attention mechanisms, particularly channel attention, have become highly influential in numerous computer vision tasks. Despite their effectiveness, many existing methods primarily focus on optimizing performance through complex attention modules applied at individual convolutional layers, often overlooking the synergistic interactions that can occur across multiple layers. In response to this gap, we introduce bridge attention, a novel approach designed to facilitate more effective integration and information flow between different convolutional layers. Our work extends the original bridge attention model (BAv1) by introducing an adaptive selection operator, which reduces information redundancy and optimizes the overall information exchange. This enhancement results in the development of BAv2, which achieves substantial performance improvements in the ImageNet classification task, obtaining Top-1 accuracies of 80.49% and 81.75% when using ResNet50 and ResNet101 as backbone networks, respectively. These results surpass the retrained baselines by 1.61% and 0.77%, respectively. Furthermore, BAv2 outperforms other existing channel attention techniques, such as the classical SENet101, exceeding its retrained performance by 0.52% Additionally, integrating BAv2 into advanced convolutional networks and vision transformers has led to significant gains in performance across a wide range of computer vision tasks, underscoring its broad applicability.