Foreground Segmentation Network with Enhanced Attention

• Published in: Shanghai jiao tong da xue xue bao 2023-06, Vol.28 (3), p.360-369
• Main Authors: Jiang, Rui, Zhu, Ruixiang, Cai, Xiaocui, Su, Hu
• Format: Article
• Language: English
• Subjects: Ablation; Architecture; Coders; Computer Science; Deep learning; Electrical Engineering; Engineering; Life Sciences; Lightweight; Materials Science; Modules; Original Paper; Robots; Segmentation; Spatial discrimination learning; Vision systems
• ISSN: 1007-1172; 1674-8115; 1995-8188
• DOI: 10.1007/s12204-023-2603-1

Moving object segmentation (MOS) is one of the essential functions of the vision system of all robots, including medical robots. Deep learning-based MOS methods, especially deep end-to-end MOS methods, are actively investigated in this field. Foreground segmentation networks (FgSegNets) are representative deep end-to-end MOS methods proposed recently. This study explores a new mechanism to improve the spatial feature learning capability of FgSegNets with relatively few brought parameters. Specifically, we propose an enhanced attention (EA) module, a parallel connection of an attention module and a lightweight enhancement module, with sequential attention and residual attention as special cases. We also propose integrating EA with FgSegNet_v2 by taking the lightweight convolutional block attention module as the attention module and plugging EA module after the two Maxpooling layers of the encoder. The derived new model is named FgSegNet_v2_EA. The ablation study verifies the effectiveness of the proposed EA module and integration strategy. The results on the CDnet2014 dataset, which depicts human activities and vehicles captured in different scenes, show that FgSegNet_v2_EA outperforms FgSegNet_v2 by 0.08% and 14.5% under the settings of scene dependent evaluation and scene independent evaluation, respectively, which indicates the positive effect of EA on improving spatial feature learning capability of FgSegNet_v2.