Image Processing Reference
In-Depth Information
Table 3
Binary Classification
Actual Positive Actual Negative
Predicted positive TP
Predicted negative FN
15. Using values in Table 3 , performance figures are generated by computing accuracy, repeat-
ability, specificity, recall, and precision metrics when classifying as fry, fingerling, and table-
ish trout. In every case, we are evaluating using as threshold from one to three standard
Accuracy, a degree of veracity, is a measurement of how well the binary classification test
correctly identifies a rainbow trout's size.
Repeatability, a degree of reproducibility, is an indicator about how robustly a rainbow trout
size can be identified:
Specificity, a degree of speciality, rates how negative rainbow trout's size is correctly identified:
Recall measures the fraction of positive examples that are correctly labelled:
Search WWH ::

Custom Search