Image recognition of complex transmission lines based on lightweight encoder-decoder networks

Table 1. Training parameters for lightweight encode-decoder network
View table
View in Article
Table 1. Training parameters for lightweight encode-decoder network
训练参数数值
Batch_size 8
Initial_lr 0.0001
Epoch 500
Cuda True

Table 2. Results of ablation experiment
View table
View in Article
Table 2. Results of ablation experiment
方法 MPA/% MIoU/% FPS
方法1 89.87 83.22 26
方法2 91.44 84.02 24
方法3 90.92 83.86 31
方法4 92.34 84.57 29
方法5 92.67 84.64 30

Table 3. Comparison of sparse training experimental results with different regularization coefficients
View table
View in Article
Table 3. Comparison of sparse training experimental results with different regularization coefficients
λ MPA/% MIoU/% Epoch
0 92.67 84.64 500
0.01 91.52 83.77 500
0.001 92.07 84.12 500
0.0001 92.58 84.56 500
0.00001 92.56 84.55 500

Table 4. Comparison of experimental results with different pruning rates
View table
View in Article
Table 4. Comparison of experimental results with different pruning rates
剪枝率 MPA/% MIoU/% FPS 参数量/(10⁶)
0 92.58 84.56 30 5.82
0.1 92.36 84.39 32 5.27
0.2 92.24 84.28 35 4.63
0.3 92.16 84.22 37 4.12
0.4 92.11 84.19 41 3.57
0.5 91.72 83.85 44 2.92

Table 5. Comparison of four network recognition results
View table
View in Article
Table 5. Comparison of four network recognition results
网络 MPA/% MIoU/% FPS 参数量/(10⁶)
PSPNet^[10] 81.86 73.78 9 178
U2Net^[6] 89.72 82.31 8 43.99
文献[7] 87.37 79.62 21 12.77
轻量型编解码网络 92.11 84.19 41 3.57

Tools

Get Citation

Copy Citation Text

Yuntang Li, Wenkai Zhu, Hengjie Li, Juan Feng, Yuan Chen, Jie Jin, Bingqing Wang, Xiaolu Li. Image recognition of complex transmission lines based on lightweight encoder-decoder networks[J]. Opto-Electronic Engineering, 2024, 51(10): 240158

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites