Abstract: The three important for every one of us is food, cloth and house. We give the first priority to food. At the same in this fast generation making a food by spending so many hours in the kitchen is a tough task. People are using various food delivery apps like Zomoto, Swiggy, Uber eats and many to get their food. Many restaurants have started takeaways courts to parcel the food items. Behind of every meal there is a procedure involved in making of it. How to the know the process behind of it. The basic of aim of this paper is to provide a solution to the query. We are introducing an inverse cooking system named it as “Swasth”. This system recreates cooking recipes for the given food image. This system provides the ingredients used and then also gives the cooking instructions. Our system uses a unique architecture to forecast ingredients as sets, modelling their relationships without enforcing any order, and then creates cooking directions while concurrently paying attention to the image and its predicted components. We thoroughly test the system on the massive Recipe 1 million dataset and demonstrate that we are able to obtain high quality recipes by utilising both image and ingredients. We also demonstrate that the system is able to produce more compelling recipes than retrieval-based approaches in terms of human judgement.
Keywords: Inverse cooking, CNN, Deep learning, Machine learning.
|
DOI:
10.17148/IJARCCE.2022.111139
