Food Recognition Using Deep Learning for Recipe and Restaurant Recommendation

Our project involves an interface that can identify various food items in real time from an image or video of multi-object meals, and provide the user with relevant information such as recipes, ingredients, and nearby restaurants serving the identified dish. Our work is divided into two parts: First, we utilize YOLO V5, an advanced object detection technique, to build a machine-learning model with an average accuracy of over 80% that can concurrently recognize and localize several items. Secondly, we integrate our model into an interface that also includes additional features such as a restaurant finder and ingredient/recipe analysis. By processing the model output on the interface side, we can display detection results in real time, including the class label and bounding box position and further recommend the recipes or the restaurants around them that serve that cuisine using an API. Hence, users can try new foods while also educating themselves about the cultures behind these cuisines.