Measuring rainbow trout by using simple statistics - Emerging Trends in Image Processing, Computer Vision, and Pattern Recognition - page 43

Table 7
Accuracy when Classifying a Probe Set at Three Standard Deviations

Testing as 1Sx (%) 2Sx (%) 3Sx (%)

Fry 81.11 80.57 69.70

Fingerling 79.11 84.41 88.86

Table-fish 71.33 81.54 96.83

Table 8
Repeatability when Classifying a Probe Set at Three Standard Deviations

Testing as 1Sx (%) 2Sx (%) 3Sx (%)

Fry 73.90 78.84 95.93

Fingerling 63.79 80.00 93.21

Table-fish 54.12 52.00 96.25

Table 9
Specificity when Classifying a Probe Set at Three Standard Deviations

Testing as 1Sx (%) 2Sx (%) 3Sx (%)

Fry 88.17 89.92 95.27

Fingerling 75.50 91.96 75.32

Table-fish 61.00 88.57 82.76

FIGURE 6 Accuracy, repeatability, and specificity performance when classifying fry, finger-
ling, and table-fish trout at three standard deviations using our statistical measuring system.

Finally, we are computing recall and precision metrics. Tables 10 and 11 summarize these
results and Figure 7 shows recall and precision results at three standard deviations.