TABLE 2.

Summary of classification methods used for the food-recognition task in image-based food-recognition systems

Classification method	Depiction	Pros	Cons
Artificial neural network (ANN)		Can achieve high recognition accuracy over 80% even when there is a nonlinear relation between the input and the output	Dependent on many parameters Low speed of computation Lack of interpretation of results
Support Vector Machine (SVM)		Can achieve high recognition accuracy over 80% even when there is a nonlinear relation between the input and the output	Binary classifiers
Naive Bayes (NB)		Can take into account prior knowledge about the domain in interest	Less accurate than other machine-learning algorithms, such as ANNs Unsuitable for large number of features
K-nearest neighbor (KNN)		Easily implemented without the need for large computational resources during execution	Sensitive to the choice of parameters
Random forest (RF)		Accuracy Speed Comprehensible by humans	—
Convolutional neural network (CNN)		Optimal results	Large datasets for training