Algorithm 2: The training algorithm of a DL model
1	Input $\to$ LC25000 dataset $DL$
2	Output $\leftarrow$ Trained model
3	BEGIN
4	Inputs $\to$ 224 × 224 × 3
5	Batch_S $\to$ 25
6	Lr $\to$ 0.0001
7	Dropout $\to$ 0.4
8	Batch_N $\to$ {momentum = 0.99, epsilon = 0.001}
9	L $\to${0,1,2,3,4}.
10	G_A_P $\to$ Global average pooling
11	Dense $\to$ = 4
12	FOR EACH image IN DL
13	Data_images. append(image)
14	END FOR
15	Splitting DL
16	Model.fc $\to$ {G_A_P, DROP, Dense}
17	Model $\to$ Model (inputs = X.inputs, outputs = Model.fc)
18	OPT $\to$ Adamx (0.0001)
19	FOR EACH lay in Model.layers [-20:]
20	IF not instance (lay, lay. Batch_N)
21	lay.trainable = True
22	END IF
23	END FOR
24	Model.compile(OPT, loss = “sparse_categorical_crossentropy”).
25	END