Deep Learning based approach to suggest recipes

This paper proposes an application that suggests recipes based on an image, clicked by the user, which contains multiple vegetables or fruits. This image can be captured in various environments, lighting conditions and from different angles. To detect multiple vegetables a state-of-art Convolutional Neural Network (CNN) called Faster-RCNN is used. Faster-RCNN is deployed on an android application which in turn interacts with server to fetch the possible recipes. The server retrieves the recipes from database, filters it and then returns the list to the client application.