Journal of Applied Optics, Volume. 45, Issue 4, 804(2024)
Electric meter data detection based on global and local multi-scale context
Figures & Tables(16)
Fig. 2. Structure diagram of convolutional module and identity module
Fig. 3. Structure diagram of local multi-scale context information extraction module
Fig. 4. Structure diagram of global context information extraction module
Table 1. Structure data of feature extraction network
View tableView in ArticleTable 1. Structure data of feature extraction network
层 输入 输出 3×3卷积层 ( 1280 , 720, 3)(640, 360, 8) 池化层 (640, 360, 8) (320, 180, 8) 卷积模块1 (320, 180, 8) (320, 180, 16) 恒等模块1 (320, 180, 16) (320, 180, 16) 卷积模块2 (320, 180, 32) (160, 90, 64) 恒等模块2 (160, 90, 64) (160, 90, 64) 卷积模块3 (160, 90, 64) (80, 45, 128) 恒等模块3 (80, 45, 128) (80, 45, 128)
Table 2. AP and MAP for 11 categories of models with different training ratios
View tableView in ArticleTable 2. AP and MAP for 11 categories of models with different training ratios
比例 类别 0 1 2 3 4 5 6 7 8 9 小数点 MAP 5∶5 0.93 0.98 0.93 0.96 0.91 0.95 0.92 0.96 0.91 0.97 0.91 0.94 6∶4 0.92 0.91 0.94 0.94 0.97 0.90 0.98 0.99 0.93 0.99 0.93 0.95 7∶3 0.95 0.98 0.96 0.93 0.94 0.94 0.99 0.95 0.96 0.97 0.94 0.96 8∶2 0.99 0.99 0.99 0.99 0.99 0.95 0.99 0.99 0.99 0.99 0.95 0.98 9∶1 0.96 0.98 0.98 0.97 0.98 0.96 0.97 0.96 0.96 0.97 0.96 0.97
Table 3. Parameter settings of comparison experiment
View tableView in ArticleTable 3. Parameter settings of comparison experiment
模型 参数 ITOIS[ 8 ]滤波方法（均值滤波），模板窗口（3，3），输入尺寸（30，0） YOLO-v3[ 9 ]输入尺寸(416，416), 主干(Darknet53), 优化器(SGD)，批次大小(8) Faster-RCNN[ 14 ]输入尺寸(600，480), 主干(Resnet50), 优化器(Adam)，批次大小(2) YOLO-v5[ 10 ]输入尺寸(640，640), 主干(CSPDarknet), 优化器(SGD)，批次大小(8) ESPA-YOLOv3[ 17 ]输入尺寸(416，416), 主干(ESPAnet), 优化器(SGD)，批次大小(8)
Table 4. AP and MAP for 11 categories of different models
View tableView in ArticleTable 4. AP and MAP for 11 categories of different models
模型 类别 0 1 2 3 4 5 ITOIS[ 8 ]0.73 0.68 0.53 0.76 0.51 0.75 YOLO-v3[ 9 ]0.82 0.71 0.64 0.61 0.87 0.80 Faster-RCNN[ 14 ]0.74 0.88 0.76 0.83 0.84 0.94 YOLO-v5[ 10 ]0.89 0.90 0.83 0.86 0.92 0.93 ESPA-YOLOv3[ 17 ]0.94 0.91 0.96 0.97 0.93 0.89 本文方法 0.99 0.99 0.99 0.99 0.99 0.95 模型 类别 6 7 8 9 小数点 MAP ITOIS[ 8 ]0.62 0.46 0.71 0.67 0.39 0.62 YOLO-v3[ 9 ]0.78 0.69 0.81 0.69 0.50 0.72 Faster-RCNN[ 14 ]0.89 0.85 0.86 0.87 0.56 0.83 YOLO-v5[ 10 ]0.86 0.96 0.93 0.95 0.71 0.89 ESPA-YOLOv3[ 17 ]0.97 0.93 0.95 0.96 0.78 0.92 本文方法 0.99 0.99 0.99 0.99 0.95 0.98
Table 5. Parameter size and running speed of different detection models
View tableView in ArticleTable 5. Parameter size and running speed of different detection models
模型 参数量/Mb FPS ITOIS[ 8 ]— 0.5 YOLO-v3[ 9 ]130 22 Faster-RCNN[ 14 ]108 5 YOLO-v5[ 10 ]122 31 ESPA-YOLOv3[ 17 ]135 20 本文模型 11.03 43
Table 6. Results of ablation experiment
View tableView in ArticleTable 6. Results of ablation experiment
全局上下文 局部上下文 MAP FPS — — 0.87 48 √ 0.92 46 √ 0.94 45 √ √ 0.98 43
Tools
Get Citation
Copy Citation Text
Tianlei MA, Jun FU, Qi MA, Zhen YANG, Xinhao LIU. Electric meter data detection based on global and local multi-scale context[J]. Journal of Applied Optics, 2024, 45(4): 804
Paper Information
Category: Research Articles
Received: Jul. 7, 2023
Accepted: --
Published Online: Oct. 21, 2024
The Author Email: FU Jun (符俊)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20