MCA: Moment Channel Attention Networks

TL;DR

This paper introduces the MCA framework, leveraging high-order statistical moments and a novel EMA mechanism to improve channel attention, leading to state-of-the-art results across multiple vision tasks with minimal computational overhead.

Contribution

The paper proposes a new Moment Channel Attention (MCA) method using extensive moment aggregation and cross moment convolution to enhance model capacity efficiently.

Findings

Achieves state-of-the-art performance on image classification, detection, and segmentation.

Outperforms existing channel attention methods with minimal additional computation.

Demonstrates the effectiveness of high-order moments in neural network attention mechanisms.

Abstract

Channel attention mechanisms endeavor to recalibrate channel weights to enhance representation abilities of networks. However, mainstream methods often rely solely on global average pooling as the feature squeezer, which significantly limits the overall potential of models. In this paper, we investigate the statistical moments of feature maps within a neural network. Our findings highlight the critical role of high-order moments in enhancing model capacity. Consequently, we introduce a flexible and comprehensive mechanism termed Extensive Moment Aggregation (EMA) to capture the global spatial context. Building upon this mechanism, we propose the Moment Channel Attention (MCA) framework, which efficiently incorporates multiple levels of moment-based information while minimizing additional computation costs through our Cross Moment Convolution (CMC) module. The CMC module via channel-wise convolution layer to capture multiple order moment information as well as cross channel features. The MCA block is designed to be lightweight and easily integrated into a variety of neural network architectures. Experimental results on classical image classification, object detection, and instance segmentation tasks demonstrate that our proposed method achieves state-of-the-art results, outperforming existing channel attention methods.