Calculation method of confusion matrix

There are three sequences in the figure below ,L Indicates the tag value ,P Represents the predicted value ,n Indicates the number of classifications .

We need to calculate n×L+P To calculate the predicted result , When L and P When taking the maximum , The result is its maximum computing space .

Such as below ,L∈(0~5),P∈(0~5), be n×L+P∈(0~35)

Then we will n×L+P Mapping to 36 One dimensional sequence of dimensions , As shown in the last sequence in the figure below , This sequence represents the calculation n×L+P The number of occurrences of each value in , So 1,2 And so on are the times of occurrence of its corresponding value .

Final , We need to get a confusion matrix , So we need to map one-dimensional sequence to two-dimensional space .

Calculate the numerical storage space of two-dimensional matrix bin=max(n×L+P)+1

Then the confusion matrix is Matrix , In the picture 0~35 The sequence of can generate 6×6 Matrix

Pictured , Press 0~35 The index of is sorted in order , Get the confusion matrix

Matrix , Blue represents line and , Green represents columns and , Pink represents diagonal elements

The evaluation index in segmentation and the calculation method of its corresponding confusion matrix are as follows