Intelligent recognition algorithm for hull segment closure surface components based on improved PointNet++

Table 1. Experimental environment configurations
View table
View in Article
Table 1. Experimental environment configurations
类别参数
系统 Ubuntu 20.04
CPU Intel i9-13900KF
GPU Nvidia GeForce RTX 4090
深度学习框架 Pytorch 1.11.0
编程语言 Python 3.9.13
显卡驱动型号 535.154.05
CUDA型号 11.3

Table 2. Input and output data formats for each layer

View table

View in Article

Table 2. Input and output data formats for each layer

网络层名	输入数据大小	输出数据大小
SA1	20×512×3	20×128×3，20×128×128
SA2	20×128×3，20×128×128	20×32×3，20×32×256
SA3	20×32×3，20×32×256	20×1×3，20×1×512
FC4	16×1×100	16×8×100
FC5	16×8×100	16×32×100
FC6	16×32×100	16×128×100
SA4	16×100×3，16×100×640	16×64×3，16×64×256
SA5	16×64×3，16×64×256	16×16×3，16×16×512
SA6	16×16×3，16×16×512	16×1×3，16×1×1024
FP6	16×1×3，16×1×1 024，16×16×3，16×16×512	16×16×512
FP5	16×16×3，16×16×512，16×64×3，16×64×256	16×64×256
FP4	16×64×3，16×64×256，16×100×3，16×100×643	16×100×512
FP3	20×1×3，20×1×512，20×32×3，20×32×256	20×32×256
FP2	20×32×3，20×32×256，20×128×3，20×128×128	20×128×128
FP1	20×128×3，20×128×128，20×512×3，20×512×3	20×512×128
FC1	20×128×512	20×64×512
DP1	20×64×512	20×64×512
FC2	20×64×512	20×16×512
DP2	20×16×512	20×16×512
FC3	20×16×512	20×4×512

Table 3. Accuracy and intersection over union （IoU）for each category
View table
View in Article
Table 3. Accuracy and intersection over union （IoU）for each category
名称板材角钢 T型材扁钢
准确率/% 94.894 88.745 82.546 80.106
交并比/% 83.831 75.341 60.839 62.043

Tools

Get Citation

Copy Citation Text

Rui LI, Yirong ZHAO, Shilin HUO, Ji WANG, Weidong SHI. Intelligent recognition algorithm for hull segment closure surface components based on improved PointNet++[J]. Chinese Journal of Ship Research, 2024, 19(6): 173

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites