FIGURE 2.

Architecture of image-based food-recognition systems for dietary assessment. The flow of the dietary assessment systems combining a camera of the user's mobile device with computer vision methods is as follows: 1) the user takes a photograph of the upcoming meal with a camera of his/her mobile device, 2) the image is preprocessed, 3) the different types of food are divided from each other through segmentation techniques, 4) robust and discriminative features are then extracted, 5) the most important features are selected and are given as input to the next phase of the system, 6) classification of food items in food categories takes place, 7) the volume of each food item is calculated, and 8) the calories and nutrients of the depicted meal are estimated using appropriate nutritional databases.