Did you ever find yourself hungry and wanted to cook something but don't know what to cook? Now you can satisfy your tummy as "Make a Recipe" is here. We plan to make an app where the user can take photos of the ingredients which the user has and the app recommends all the possible dishes which can be made with those ingredients by recognizing them.
- Django - Backend Server.
- Android - Client Side.
- CNN using Keras - For recognising the ingredients in the images.
- Web Scraping using bs4(Beautiful soup)- For scraping the images and required text from web and fetching the reicpes.
Download the apk file(Make A Recipe) from this repo and install the downloded apk file. Since the installing apk file is not from any store please enable the permissions to install the apk file. The android version of the app is kept low to enable it for maximum number of users the necessary permissions are asked only when they are required.
https://drive.google.com/open?id=1vS3RqytLVNh9tjx9Z--Xsv1XKCPOhxJN
- The user opens his/her app and takes a photo of an ingredient and hits the submit button.
- The image is then sent to backend which is written with Django and deployed to Heroku.
- The server takes the image, predicts what ingredient is present in the image witht the help of CNN and sends it back to the client.
- The client can verify it and he/she will be asked if there are more ingredients to take photos of.
- If the answer is yes, steps 2 to 4 are repeated.
- If answer is no, The list of all the ingredients which are detected will be shown on screen.
- After that, The scraping algorithms will come in and searches the web for the dishes that can be made with the ingredients detected in the earlier step.
- Two of the top most popular dishes will be shown.
- The app will also recommend, the best combination for each of the displayed item. ex: if you have all the ingredients to make Biryani. It would suggest you to add Kuchember to it.
- The app is not complete if it doesn't show you how to make the recommended dishes. It shows you some youtube videos on how to make the recipes which are possible with the ingredients you have.
- The image processing part wasnt complete with all the items, it was trained only with minimal no.of items(in consideration of Hackathon). 12.The usage of necessary data for image processing part is less compared to the real world problems, for increasing the accuracy more data should be used.
- Django is an open source software.
- Android is free to use for every user.
- Web scrapping using Beautifulsoup to collect data and information and its free to use and it is licensed under MIT.
- Keras for image detection and its free to use and it is licensed under MIT.
- Used allrecipies site and here is permission for the user "Allrecipes.com's apps for smartphones allow users to access the site and its user-uploaded content while on the go"