Optics and Precision Engineering, Volume. 31, Issue 2, 277(2023)
Real time semantic segmentation network of wire harness terminals based on multiple receptive field attention
Figures & Tables(11)
Table 1. Module performance index comparison
View tableView in ArticleTable 1. Module performance index comparison
Method K MIoU/% MPA/% Dice/% FPS/frame UNet — 96.94 98.51 97.87 5 STDC — 96.77 98.53 97.72 20 MRF-UNet 32 97.22 98.70 98.08 15 16 97.54 98.83 98.31 15 8 97.30 98.69 98.14 15 4 96.88 98.57 97.81 15 2 96.95 98.62 97.86 14
Table 2. Comparison of network performance indicators
View tableView in ArticleTable 2. Comparison of network performance indicators
Net Memory/MB FLOPs/G FPS/frame UNet 2 673.63 282.07 5 SegNet 1 580.27 282.50 7 BiSeNet 1 638.56 77.16 12 MRF-UNet 1 397.48 31.47 15
Table 3. Comparison of network segmentation effects
View tableView in ArticleTable 3. Comparison of network segmentation effects
Net MIoU MPA Dice UNet 96.94 98.51 97.87 SegNet 96.71 98.52 97.68 BiSeNet 94.91 98.00 96.24 MRF-UNet 97.54 98.83 98.31
Tools
Get Citation
Copy Citation Text
Yanan GU, Ruyi CAO, Lishan ZHAO, Bibo LU, Baishun SU. Real time semantic segmentation network of wire harness terminals based on multiple receptive field attention[J]. Optics and Precision Engineering, 2023, 31(2): 277
Paper Information
Category: Information Sciences
Received: Jun. 13, 2022
Accepted: --
Published Online: Feb. 9, 2023
The Author Email: SU Baishun (subaishun@hpu.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20