System for the automatic separation of documents in a batch of documents

A method for separating documents in a batch of unseparated documents, the method comprising:

• receiving, as input, the batch of unseparated documents;
• applying, with an electronic processor, image processing to each page in the batch of documents;
• determining, for each pair of consecutive pages in the batch of documents, a probability that pages of the pair of consecutive pages belong to different documents using a predictive model;
• generating, with the electronic processor, a batch of separated documents by providing an indication of a document boundary if the probability that pages of a pair of consecutive pages belong to different documents is above a predetermined threshold;
• and
• performing at least one selected from a group of displaying, via a display, the batch of separated documents, storing, in a memory of the electronic processor, the batch of separated documents, and transmitting the batch of separated documents;
• wherein the predictive model is trained using a training set of pairs of pages and a first percent of the pairs of pages are pages that belong to the same document, a second percent of the pairs of pages are pages that belong to different documents that are from a single source, and a remaining percent of the of the pairs of pages are pages that belong to different documents that are from different sources.

A system for separating documents in a batch of unseparated documents, the system comprising:

• a display;
• and
• an electronic processor electrically connected to the display and configured to
• receive, as input, the batch of unseparated documents;
• apply, image processing to each page in the batch of documents;
• determine, for each pair of consecutive pages in the batch of documents, a probability that pages of the pair of consecutive pages belong to different documents using a convolutional network;
• generate a batch of separated documents by providing an indication of a document boundary if the probability that pages of a pair of consecutive pages belong to different documents is above a predetermined threshold;
• and
• perform at least one selected from a group of displaying, via the display, the batch of separated documents, storing, in a memory of the electronic processor, the batch of separated documents, and transmitting the batch of separated documents.