A feature refinement module for light-weight semantic segmentation network

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Wang, Zhiyan, Guo, Xin, Wang, Song, Zheng, Peixiao, Qi, Lin
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910741991260160
author Wang, Zhiyan
Guo, Xin
Wang, Song
Zheng, Peixiao
Qi, Lin
author_facet Wang, Zhiyan
Guo, Xin
Wang, Song
Zheng, Peixiao
Qi, Lin
contents Low computational complexity and high segmentation accuracy are both essential to the real-world semantic segmentation tasks. However, to speed up the model inference, most existing approaches tend to design light-weight networks with a very limited number of parameters, leading to a considerable degradation in accuracy due to the decrease of the representation ability of the networks. To solve the problem, this paper proposes a novel semantic segmentation method to improve the capacity of obtaining semantic information for the light-weight network. Specifically, a feature refinement module (FRM) is proposed to extract semantics from multi-stage feature maps generated by the backbone and capture non-local contextual information by utilizing a transformer block. On Cityscapes and Bdd100K datasets, the experimental results demonstrate that the proposed method achieves a promising trade-off between accuracy and computational cost, especially for Cityscapes test set where 80.4% mIoU is achieved and only 214.82 GFLOPs are required.
format Preprint
id arxiv_https___arxiv_org_abs_2412_08670
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A feature refinement module for light-weight semantic segmentation network
Wang, Zhiyan
Guo, Xin
Wang, Song
Zheng, Peixiao
Qi, Lin
Computer Vision and Pattern Recognition
Machine Learning
Image and Video Processing
Low computational complexity and high segmentation accuracy are both essential to the real-world semantic segmentation tasks. However, to speed up the model inference, most existing approaches tend to design light-weight networks with a very limited number of parameters, leading to a considerable degradation in accuracy due to the decrease of the representation ability of the networks. To solve the problem, this paper proposes a novel semantic segmentation method to improve the capacity of obtaining semantic information for the light-weight network. Specifically, a feature refinement module (FRM) is proposed to extract semantics from multi-stage feature maps generated by the backbone and capture non-local contextual information by utilizing a transformer block. On Cityscapes and Bdd100K datasets, the experimental results demonstrate that the proposed method achieves a promising trade-off between accuracy and computational cost, especially for Cityscapes test set where 80.4% mIoU is achieved and only 214.82 GFLOPs are required.
title A feature refinement module for light-weight semantic segmentation network
topic Computer Vision and Pattern Recognition
Machine Learning
Image and Video Processing
url https://arxiv.org/abs/2412.08670