In this project introduces an inverse cooking system that takes a food image and recreate a cooking recipes consisting of a title, ingredients, sequence of cooking instructions, Calorie, Cooking time.
This model predicts ingredients as sets by means of DenseNet201 and Recipe Retrieval Algorithm. Generates cooking instructions by attending to both image and its inferred ingredients simultaneously.
The recipes from the website are scraped and saved as a JSON file, google_images_download library is used to download images. For each recipe 10 images are downloaded for training and 2 for testing.
A total of 9432 images are downloaded (7860 train and 1572 test)
There is no real time web portal available to predict the food based on uploaded image. Currently entertainment portal like YouTube are used to find the recipe of a food item but it only works while we know the name of the food item.
