Grocery Receipt Data Extraction AI
Budget: €30 – €250 EUR
I have a project involving the extraction of data from grocery receipts. These receipts have been photographed and saved as JPEGs. Some receipts are straightforward while others may be a bit more complex due to their varying structures.
The AI I am looking for should accurately retrieve the following information from these receipts:
- Supermarket details: Name and Address
- Receipt specifics: Date, Total, and Total net (if available)
- Tax items: For each tax-rate, the AI should identify: 1) net amount 2) tax amount 3) tax code (A or B) 4) tax rate 5) total
- Line items: Each item on the receipt should be described in terms of: 1) Item description 2) Quantity 3) Price per piece 4) Total on the line item
My previous attempts with Google Cloud Document AI + Invoice Extractor were not entirely successful, as they don't cater specifically to supermarket receipts.
I am open to your proposals which may include:
1) Labelling in Google Cloud with my files followed by the generation of a new custom model that performs over 90% accuracy for line items and over 98% for other fields,
2) Or any other innovative solution you may have.
Ideal candidates for this project should have experience with AI and data extraction, particularly from receipt images. Please note that the receipt images are JPEGs and the main challenge lies in the variant structures of these grocery receipts.
The AI I am looking for should accurately retrieve the following information from these receipts:
- Supermarket details: Name and Address
- Receipt specifics: Date, Total, and Total net (if available)
- Tax items: For each tax-rate, the AI should identify: 1) net amount 2) tax amount 3) tax code (A or B) 4) tax rate 5) total
- Line items: Each item on the receipt should be described in terms of: 1) Item description 2) Quantity 3) Price per piece 4) Total on the line item
My previous attempts with Google Cloud Document AI + Invoice Extractor were not entirely successful, as they don't cater specifically to supermarket receipts.
I am open to your proposals which may include:
1) Labelling in Google Cloud with my files followed by the generation of a new custom model that performs over 90% accuracy for line items and over 98% for other fields,
2) Or any other innovative solution you may have.
Ideal candidates for this project should have experience with AI and data extraction, particularly from receipt images. Please note that the receipt images are JPEGs and the main challenge lies in the variant structures of these grocery receipts.