Introduction

Repeat Image Check provides the user the capability to check whether the document image being submitted has been seen before or not. This can be either a publicly available image or re-submitting a document which has been processed by the client in the past.

An input document can be in form of an image or PDF. If a PDF is submitted, the first page of the PDF is converted into an image and used for further processing.

The input document image is converted into a vector representation of itself using an embedder. This vector is compared to a historical dataset of embedded documents. A similarity score is calculated between the input image and historic images. Based on preset thresholds, similarity matches are returned.

Solution Flow