Human Object Classification Using Dual Tree Complex Wavelet Transform and Zernike Moment - Transactions on Large-Scale Data-and Knowledge-Centered Systems - page 97

Table 4. Confusion matrix for the proposed method using discrete wavelet transform (DWT)
coefficients at multilevel as a feature set

DWT coefficients (Level - 1) as
a feature set

DWT coefficients (Level - 2) as
a feature set

Predicted classes Predicted classes

True classes True Classes

Human Non-human Human Non-human

Human 85 15 Human 85 15

Non-human 30 70 Non-human 30 70

DWT coefficients (Level - 3) as
a feature set

DWT coefficients (Level - 4) as
a feature set

Predicted classes Predicted classes

True classes True classes

Human Non-human Human Non-human

Human 89 11 Human 91 9

Non-human 27 73 Non-human 21 79

DWT coefficients (Level - 3) as
a feature set

DWT coefficients (Level - 4) as
a feature set

Predicted classes Predicted classes

True classes True classes

Human Non-human Human Non-human

Human 93 7 Human 95 5

Non-human 20 80 Non-human 11 89

DWT coefficients (Level - 7) as a feature set

Predicted classes

True classes

Human Non-human

Human 95 5

Non-human 11 89

Table 5. Confusion matrix for the state-of-the-art methods [7,9,14,15]

Method proposed by
Dalal and Triggs [7]

Method proposed by Lu and Zheng [9]

Predicted classes Predicted classes

True classes True Classes

Human Non-human Human Non-human

Human 96 04 Human 90 10

Non-human 08 92 Non-human 30 70

Method proposed by Renno et al. [14] Method proposed by Chen et al. [15]

Predicted classes Predicted classes

True classes True classes

Human Non-human Human Non-human

Human 89 11 Human 98 02

Non-human 31 69 Non-human 04 96

Average classification accuracy is defined as the proportion of the total number of
prediction that were correct. Average classification accuracy can be calculated using
following formula.

TP + TN

Average Classification Accuracy =

(8)

TP +++ TN FP FN