MFPNet: Multi-scale Feature Propagation Network For Lightweight Semantic Segmentation

TL;DR

MFPNet is a lightweight semantic segmentation model that enhances feature representation by using multi-scale context exploration and graph convolutional networks for feature propagation.

Contribution

The paper introduces MFPNet, a novel lightweight architecture with a symmetrical encoder-decoder structure and GCN-based feature propagation for improved segmentation.

Findings

Achieves superior results on benchmark datasets

Effectively models long-range contextual relationships

Utilizes multi-scale feature propagation for better accuracy

Abstract

In contrast to the abundant research focusing on large-scale models, the progress in lightweight semantic segmentation appears to be advancing at a comparatively slower pace. However, existing compact methods often suffer from limited feature representation capability due to the shallowness of their networks. In this paper, we propose a novel lightweight segmentation architecture, called Multi-scale Feature Propagation Network (MFPNet), to address the dilemma. Specifically, we design a robust Encoder-Decoder structure featuring symmetrical residual blocks that consist of flexible bottleneck residual modules (BRMs) to explore deep and rich muti-scale semantic context. Furthermore, taking benefit from their capacity to model latent long-range contextual relationships, we leverage Graph Convolutional Networks (GCNs) to facilitate multi-scale feature propagation between the BRM blocks. When evaluated on benchmark datasets, our proposed approach shows superior segmentation results.