Infrared Target Imaging Liquid Level Detection Method Based on Deep Learning

Fig. 7. Visualization of the training process. (a) Box loss; (b) Objectness loss; (c) Classification loss; (d) validation set Box loss; (e) validation set Objectness loss; (f) validation set Classification loss; (g) precision; (h) recall; (i) mAP

Download full size

Fig. 8. Test results of liquid level state

Download full size

Table 1. Performance indicators and calculation methods of the evaluation model

View table

Table 1. Performance indicators and calculation methods of the evaluation model

Indicator	Describe	Calculation method
Accuracy	ratio of correctly classified samples to total samples	$\frac{X_{TP} + X_{TN}}{X_{TP} + X_{FN} + X_{FP} + X_{TN}}$
Percision (P)	ratio of correctly retrieved samples to total retrieved samples	$\frac{X_{TP}}{X_{TP} + X_{FP}}$
Recall (R)	ratio of correctly retrieved samples to samples that should have been retrieved	$\frac{X_{TP}}{X_{TP} + X_{FN}}$
F-Score	consider the reconciliation between accuracy and recall	$\frac{(1 + β^{2}) \cdot P \cdot R}{β^{2} \cdot (P + R)}$

Table 2. Positive and negative sample relationship
View table
Table 2. Positive and negative sample relationship
Case Test outcome Truth
X_TP anomaly anomaly
X_FP anomaly normol
X_FN normol anomaly
X_TN normal normal

Tools

Get Citation

Copy Citation Text

Xiao Liang, Jiawei Li, Xiaolong Zhao, Junbin Zang, Zhidong Zhang, Chenyang Xue. Infrared Target Imaging Liquid Level Detection Method Based on Deep Learning[J]. Acta Optica Sinica, 2021, 41(21): 2110001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites