WAY2VAT Receives Patent for Deep Learning Image Tagging Tool
New technology expected to improve invoice analysis accuracy
October 15, 2020 – WAY2VAT, the leader in automated VAT reclaim for business, was awarded a patent for its groundbreaking VAT analysis services. The patent describes a deep learning method with convolutional neural networks that combine image and text to analyze invoice images and extract field data to recover the transaction information. Using this technology, the deep learning tools are capable of reading scanned data more accurately than existing models on the market today.
Automated VAT reclaim solutions typically scan invoices and receipts. However, automated systems struggle to identify information on slips of paper that are frequently crumpled, have printing imperfections, missing data, and printed using low-quality dot matrix fonts on paper that fades quickly. WAY2VAT’s new patent uses a series of unique process called Skip-Rect Embedding, OCR Error Correction with Skip-Rect Embedding, and Deep Field tagging to fully extract usable data.
As a result, businesses are able to use WAY2VAT’s automated processes to recover larger amounts of VAT. The deep learning elements within the system is capable of recognizing and understanding data, which is used as part of the process in VAT recovery.
This patent is one of several that WAY2VAT has been awarded. As the first vendor with patented AI technology WAY2VAT is able to increase the amount of reclaimed VATby as much as 30%. WAY2VAT looks forward to continuing to serve its global customers and facilitating refunds which can be as high as 27% of any invoice.