Global Aggregation then Local Distribution in Fully Convolutional Networks

TL;DR

GALD enhances fully convolutional networks by combining global aggregation with local distribution, improving boundary and small object recognition, and achieving state-of-the-art results in semantic segmentation.

Contribution

The paper introduces GALD, a novel module that sequentially applies global aggregation and local distribution, improving feature modeling in FCNs for various vision tasks.

Findings

Achieves state-of-the-art mIoU 83.3% on Cityscapes.

Consistently improves performance of existing FCN-based methods.

End-to-end trainable and easily integrable into existing architectures.

Abstract

It has been widely proven that modelling long-range dependencies in fully convolutional networks (FCNs) via global aggregation modules is critical for complex scene understanding tasks such as semantic segmentation and object detection. However, global aggregation is often dominated by features of large patterns and tends to oversmooth regions that contain small patterns (e.g., boundaries and small objects). To resolve this problem, we propose to first use \emph{Global Aggregation} and then \emph{Local Distribution}, which is called GALD, where long-range dependencies are more confidently used inside large pattern regions and vice versa. The size of each pattern at each position is estimated in the network as a per-channel mask map. GALD is end-to-end trainable and can be easily plugged into existing FCNs with various global aggregation modules for a wide range of vision tasks, and consistently improves the performance of state-of-the-art object detection and instance segmentation approaches. In particular, GALD used in semantic segmentation achieves new state-of-the-art performance on Cityscapes test set with mIoU 83.3\%. Code is available at: \url{https://github.com/lxtGH/GALD-Net}