ReAFFPN: Rotation-equivariant Attention Feature Fusion Pyramid Networks for Aerial Object Detection

TL;DR

ReAFFPN introduces a rotation-equivariant attention mechanism within a feature fusion pyramid network to enhance aerial object detection accuracy while maintaining rotation invariance.

Contribution

It proposes a novel rotation-equivariant channel attention method integrated into a feature fusion network for improved aerial object detection.

Findings

Achieves better rotation-equivariant feature fusion

Significantly improves detection accuracy

Maintains rotation invariance in features

Abstract

This paper proposes a Rotation-equivariant Attention Feature Fusion Pyramid Networks for Aerial Object Detection named ReAFFPN. ReAFFPN aims at improving the effect of rotation-equivariant features fusion between adjacent layers which suffers from the semantic and scale discontinuity. Due to the particularity of rotational equivariant convolution, general methods are unable to achieve their original effect while ensuring rotation equivariance of the network. To solve this problem, we design a new Rotation-equivariant Channel Attention which has the ability to both generate channel attention and keep rotation equivariance. Then we embed a new channel attention function into Iterative Attentional Feature Fusion (iAFF) module to realize Rotation-equivariant Attention Feature Fusion. Experimental results demonstrate that ReAFFPN achieves a better rotation-equivariant feature fusion ability and significantly improve the accuracy of the Rotation-equivariant Convolutional Networks.