Image based drinks identification for dietary assessment

In this paper, we propose a method to detect and recognize different types of drinks with a view to estimate its nutrient values using vision based algorithms. We use visual saliency and thresholding techniques to segment out the drink region from the image. After detection, both speed up robust features and color based visual features are used to create a Bag of Features (BoF). This BoF model is employed for recognition of drinks. Based on experimentation, our proposed method confirms that it is capable of detecting and recognizing different drinks categories with an accuracy greater than 89%.