Culinary Fusion:Vegetable Recognition Recipe Generation

The presented paper aims to propose a novel way to implement advanced deep learning and natural language processing in vegetable recognition and recipe generation. The proposed work uses the YOLOv8 object detection framework, which is first trained on a very meticulously obtained dataset of 20, 708 images from the website Roboflow, to perform intelligent partitioning into three sets: Finally, the proposed methodology has attained 88% accuracy in predicting different vegetables. On detecting vegetables in any input image, the model uses OpenCV to detect those vegetables from the PALM API. It takes the names of the detected vegetables as prompts along with the preferences selected by the user in terms of the language and the number of recipes and gives the output recipe using the PALM API in the form of five different language translations, Telugu, Malayalam, Hindi, Kannada, and Tamil. The output recipes are also provided in text and audio format to cater to users of all linguistic backgrounds. This results in a holistic system that ensures a hassle-free and user-friendly approach to vegetable recognition, its recipe generation, and, finally, the enrichment of the culinary experience and promotion of cultural diversity, even in the kitchen.