Mandibular fracture detection with 3M-YOLOv5 network based on enhanced feature extraction capability

网络	参数量/MB	GFLOPS	精确率	召回率	F1值	mAP值
YOLOv5	7.064	16.477	0.953 8	0.948 9	0.953 1	0.956 1
实验一	12.220	70.187	0.986 2	0.971 0	0.978 5	0.990 9
实验二	9.997	58.818	0.974 3	0.967 6	0.970 9	0.982 2
实验三	8.924	54.124	0.982 8	0.972 7	0.977 7	0.984 1
实验四	10.251	61.116	0.982 9	0.977 9	0.980 4	0.989 0
实验五	9.389	52.430	0.982 8	0.971 0	0.976 9	0.987 8
实验六	8.780	46.764	0.973 3	0.930 2	0.951 2	0.977 3
实验七	11.966	67.862	0.986 0	0.957 4	0.971 5	0.979 2
实验八	8.924	54.098	0.974 4	0.971 0	0.972 7	0.979 1
实验九	11.104	59.675	0.881 6	0.913 1	0.897 1	0.920 8
实验十	10.251	61.143	0.993 2	0.988 1	0.990 6	0.991 7

Table 2. Comparison of experiment results with different networks

View table

View in Article

Table 2. Comparison of experiment results with different networks

网络	参数量/MB	GFLOPS	精确率	召回率	F1值	mAP值	FPS
YOLOv3^［19］	236.32	66.171	0.844 6	0.277 7	0.417 9	0.638 9	11.819 3
YOLOv4^［20］	245.53	60.527	0.908 3	0.691 7	0.785 3	0.851 5	9.465 7
YOLOv5	7.064	16.477	0.953 8	0.948 9	0.953 1	0.956 1	14.051 6
Faster RCNN^［21］	137.099	370.210	0.919 7	0.858 6	0.888 1	0.927 4	7.059 4
CenterNet^［22］	32.665	70.217	0.877 5	0.524 7	0.656 7	0.794 0	24.972 8
YOLOX^［23］	8.938	26.635	0.918 4	0.901 2	0.909 7	0.911 8	22.099 2
YOLOv7^［24］	37.195	104.761	0.910 3	0.812 6	0.858 7	0.901 6	12.669 1
3M-YOLOv5	10.251	61.143	0.993 2	0.988 1	0.990 6	0.991 7	2.682 6

Tools

Get Citation

Copy Citation Text

Tao ZHOU, Yuhu DU, Daozong SHI, Caiyue PENG, Huiling LU. Mandibular fracture detection with 3M-YOLOv5 network based on enhanced feature extraction capability[J]. Optics and Precision Engineering, 2023, 31(21): 3178

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites