Algorithm 3: DL_Models_Service

Input: skin_image_dataset Output: tf_model, tf_lite_model files

1   Function: design_models (skin_image_dataset) 2    Init: size ← input dimension, std ← normalization factor 3    tf_file_name ← “TF_model”, tf_lite_file_name ← “TFLite_model”     //Model designing 4    tf_model ← design_tf_model()     //Model training 5     For skin_image in skin_image_dataset        //Image pre-processing 6      skin_image ← skin_image.rsize(size, size)//Resize the image to size x size 7      img_array[size, size] ← convert_to_array(skin_image)//Convert image to an array         //Normalization 8      For x in img_array 9        For y in img_array[x] 10           img_array[x, y] ← img_array[x, y]/std 11         End For 12       End For 13       tf_model.train_model(img_array) 14    End For//end training         //save TF model 15    tf_model.save_model(tf_file_name)         //Convert to TFLite model 16    tf_lite_model ← convert_to_tflite(tf_model) 17    tf_lite_model.save_model(tf_lite_file_name)