We have thousands of different images of different types of receipts that are all taken in different lighting and angles etc. We need some software written using PHP and MySQL (for db if needed) that will do any necessary pre processing and resizing to the image to allow for best results for use Google Vision API to extract receipt information from a receipt image and from this extract the following entities from the receipt data: Merchant Name, Date, Total, Tax, Payment Method (cash, credit, or debit), and last 4 digits of payment method.
We have many different types of receipts and new ones coming in constantly, so we want this code to use machine learning to classify the receipts and each piece of information so that it is always learning as new receipts are presented in order to achieve best results.
I would like for the end product to allow us to make a simple post to a php page that includes an image of a receipt and it will return a json object with what it has determined for the entities confidence level in percentage.
{
'merchant_name': 'Walmart',
'total': '100.00',
'tax': '5.00',
'date': '1970-05-12',
'payment_method': 'CREDIT',
'last_four':'4556',
'confidence_level', '0.73'
}
About the recuiterMember since May 20, 2018 Ooha Paruchuri
from Ontario, Canada