Table 4. Optimal parameters used for machine learning and CNN models.

Algorithms	Optimal parameter
Algorithms	Used Parameters
Random Forest Classifier	n_estimators = 10 to 500, max_depth = 10 to 30, min_samples_split = 10,random_state = 45,class_weight = ‘balance’
K-Nearest Neighbors (KNN)	neighbors = 5, metric = ’minkowski’ p = 4, weights = ’distance’, algorithm = ’auto’ or ’ball_tree’
AdBoostClassifier	n_estimators = 200 to 500, learning_rate = 0.06 to 0.1, max_depth = 3 to 5
Neural Network	hidden_layer_sizes = (32,64,128,256,1024), activation = ’relu’, solver = ’adam’, alpha = 0.0001, learning_rate = ’adaptive’, Dropout(0.2), loss = ’binary_crossentropy’, epochs = 100, batch_size = 10
CNN	Image size = 224*224*3, 70% training and 30% testing, adaptive moment estimation (Adam) optimizer, mini patch size of 32, maximal epochs of 60, and initial learning rate of 0.001.