Algorithm 1. Stack Ensemble

Input: Segmented images ($\propto$) = {(xi, yi) | xi $\in$ X, yi $\in \mathit{Y}$} Output: Ensemble classifier (Ec)         Step 1: Train base - models ($\mathit{\phi }$) from segmented images ($\propto$)         For p$\leftarrow$ 1 P do                 Train base model $\mathit{\phi }$ base on $\propto$                 Aggregate obtained predictions from all 8 base models         Step 2: Create a new dataset ($\mathit{\beta }$) from base model predictions.         For n$\leftarrow$ 1 to z do                 Create new dataset comprising {xi, yi}, where xi = {$\mathit{\beta }$j (xi) for j = 1 − 8}         Step 3: Train a second-level meta learner                 Learn a new classifier Ec based on newly created dataset         Return Es(x) = es(es1(x), es2(x)… es8 (x))