Journal of Computer Applications ›› 2023, Vol. 43 ›› Issue (3): 936-942.DOI: 10.11772/j.issn.1001-9081.2022020210

• Multimedia computing and computer simulation • Previous Articles    

DeepLabV3+ image segmentation algorithm fusing cumulative distribution function and channel attention mechanism

Xuedong HE1, Shibin XUAN1,2(), Kuan WANG1, Mengnan CHEN1   

  1. 1.School of Artificial Intelligence,Guangxi Minzu University,Nanning Guangxi 530006,China
    2.Guangxi Key Laboratory of Hybrid Computation and IC Design and Analysis (Guangxi Minzu University),Nanning Guangxi 530006,China
  • Received:2022-02-24 Revised:2022-05-25 Accepted:2022-05-25 Online:2022-08-16 Published:2023-03-10
  • Contact: Shibin XUAN
  • About author:HE Xuedong, born in 1997, M. S. candidate. His research interests include semantic segmentation, computer vision.
    WANG Kuan, born in 1995, M. S. candidate. His research interests include pose estimation, deep learning.
    CHEN Mengnan, born in 1997, M. S. candidate. His research interests include arithmetic algorithm optimization, computational intelligence.
  • Supported by:
    National Natural Science Foundation of China(61866003)

融合累积分布函数和通道注意力机制的DeepLabV3+图像分割算法

何雪东1, 宣士斌1,2(), 王款1, 陈梦楠1   

  1. 1.广西民族大学 人工智能学院,南宁 530006
    2.广西混杂计算与集成电路设计分析重点实验室(广西民族大学),南宁 530006
  • 通讯作者: 宣士斌
  • 作者简介:何雪东(1997—),男,吉林松原人,硕士研究生,CCF会员,主要研究方向:语义分割、计算机视觉
    宣士斌(1964—),男,安徽无为人,教授,博士,主要研究方向:图像处理与识别
    王款(1995—),男,江苏海安人,硕士研究生,主要研究方向:姿态估计、深度学习
    陈梦楠(1997—),男,山西长治人,硕士研究生,主要研究方向:算法优化、计算智能。
  • 基金资助:
    国家自然科学基金资助项目(61866003)

Abstract:

In order to solve the problems that the low-level features of the backbone are not fully utilized, and the effective features are lost due to large-times upsampling in DeepLabV3+ semantic segmentation, a Cumulative Distribution Channel Attention DeepLabV3+ (CDCA-DLV3+) model was proposed. Firstly, a Cumulative Distribution Channel Attention (CDCA) was proposed based on the cumulative distribution function and channel attention. Then, the cumulative distribution channel attention was used to obtain the effective low-level features of the backbone part. Finally, the Feature Pyramid Network (FPN) was adopted for feature fusion and gradual upsampling to avoid the feature loss caused by large-times upsampling. On validation set Pascal Visual Object Classes (VOC)2012 and dataset Cityscapes, the mean Intersection over Union (mIoU) of CDCA-DLV3+ model was 80.09% and 80.11% respectively, which was 1.24 percentage points and 1.02 percentage points higher than that of DeepLabV3+ model. Experimental results show that the proposed model has more accurate segmentation results.

Key words: deep learning, image semantic segmentation, channel attention mechanism, DeepLabV3+, cumulative distribution function

摘要:

为了解决DeepLabV3+在语义分割时未充分利用主干的低级特征,以及大倍数上采样造成有效特征缺失的问题,提出一种累积分布通道注意力DeepLabV3+(CDCA-DLV3+)模型。首先,基于累积分布函数和通道注意力提出了累积分布通道注意力(CDCA);然后,利用CDCA获取主干部分有效的低级特征;最后,采用特征金字塔网络(FPN)进行特征融合和逐步上采样,从而避免大倍数上采样所造成的特征损失。CDCA-DLV3+模型在Pascal VOC 2012验证集与Cityscapes数据集上的平均交并比(mIoU)分别为80.09%和80.11%,相较于DeepLabV3+模型分别提升1.24和1.02个百分点。实验结果表明,所提模型分割结果更加精确。

关键词: 深度学习, 图像语义分割, 通道注意力机制, DeepLabV3+, 累积分布函数

CLC Number: