Many vehicle owners are probably quite familiar with the European Accident Statement (EAS) form. Although, if you have not come across it before, it may be best to keep it that way. This is because the EAS form is the European standardised form you have to fill if you get into a car or bike accident and want to file your insurance claim.
Paperbox is one of the world’s foremost Insurance Document Processors (IDP). The EAS form is a document that is quite popular in many insurers’ Property and Casualty (P&C) Claims document flows. Therefore, you can expect that we encounter this particular document quite commonly. The EAS form is usually completed in highly stressful conditions – understandably because it appears on the heels of a road traffic accident. Combine this state with a poor stylo or pencil, and an “edge gase” quickly results. In the Insurance Document Processors domain, an “edge case” is a document that is extremely difficult to be automatically recognized and extracted. At Paperbox, this is exactly the type of documents we are made for. In this article, our Product Lead Maarten Callaert shares his vision on how Paperbox succeeds in extracting EAS forms. Read on. We get around them by the unique combination of three different modules.
The Paperbox engine consists of Three Phases: Digitization, Document Classification and Information Extraction. The challenge posed by an European Accident Statement form is usually on Digitization and Extraction. Let’s start with Extraction. Before Digitization (done by deploying an Optical Character Recognition (OCR) engine), we split the EAS form by activating a Paperbox Computer Vision model. This model, which is shown below, allows us to divide the document into boxes (for example, Party A > Car). Furthermore, you should know that the same model is used to Extract (and even Classify) checkboxes, situation drawings and signatures. In addition to digitizing the handwritten annotations on the document, Paperbox uses an extensive suite of Optical Character Recognition software to convert the scans or handwritten annotations into fully digitized formats.
With this, we achieve decent accuracy even in these handwritten situations. However, we are the first to admit that this isn’t usually entirely accurate. To make up for this, the other two components come into play.
The Paperbox Product and User Interface are designed to exactly handle documents like this. For one, it allows the Property and Casualty claims handler to use different features to review, validate and correct such documents. For example, the Paperbox hotkeys enable our customers to move quickly to the next page (using ←/→) or to the next form (with Ctrl + ←/→). In addition, the User Interface makes it also possible to consult and review the visual extraction elements like checkboxes and situation drawing. Also, the OCR processing happens on top of the EAS form. If a particular field has not been extracted successfully, the P&C claims handler can easily select a value by highlighting the undetected data element.
Paperbox is a Software-as-a-Service solution in the insurance industry with a focus on converting unstructured documents and communications into structured data. The Paperbox solution is equipped with appropriate integration features to provide and then check the extracted elements with the insurer’s downstream (claims) systems. In addition to the above technologies, our Engineers always start a project by challenging the status quo and spot some quick wins. For example, it is unnecessary to extract all 200 fields of the European Accident Statement form when you could more easily look up particular fields using a unique identifier, especially for a Party insured in your business. There you have it! This is what extracting a European Accident Statement form at Paperbox entails. If you are interested in seeing Paperbox in action at your Property and Casualty Claims document flows or with your EAS forms, you can send an email to email@example.com