Novel insurance policy identification instance segmentation method

Abstract

The invention provides a novel insurance policy identification instance segmentation method, and the method mainly comprises two parts: 1, carrying out the line detection analysis of a calculation result based on the characteristics of an insurance policy general DeepLabV3 neural network model, carrying out the clustering analysis of a detection result of a boundary line of each instance on the basis of a general DeepLabV3 algorithm result, carrying out secondary calculation and inference on the attribute of each instance; and 2, correcting the instance, so that the inclined insurance policy instance is corrected into an insurance policy instance which is horizontal and vertical. According to the method, the instance segmentation of the insurance policy picture can be automatically realized, the insurance policy part and the invalid background information can be found, the invalid background information can be automatically deleted, and more accurate and concise picture information can be provided for OCR (Optical Character Recognition) operation. Through testing of more than 200,000 insurance policies, the method improves the insurance policy identification accuracy by 27.2%.

Description

technical field [0001] The invention relates to the technical field of insurance policy recognition instance segmentation, and proposes a novel insurance policy recognition instance segmentation method named as QZG-DeepLab (QuanZhangGui-DeepLab). Background technique [0002] Policy recognition refers to a technical solution for converting paper policy into unstructured plain text through OCR technology, which can be invoked by other business processes in the insurance industry. With the rapid development of my country's economy and the substantial increase in the level of national income, the number of insurance policies held by the public is also soaring rapidly, which puts forward higher requirements for the digital management level of the insurance industry. As of 2020, there are a total of 1.67 billion insurance policies in my country, with an average of 1.19 policies per capita. Among them, there are less than 190 million digital insurance policies, and the vast major...

AI Technical Summary

Problems solved by technology

Among them, there are less than 190 million digital insurance policies, and the vast majority of insurance policies are still paper policies, which have not been digitized, causing inconvenience for customers in many links such as claim settlement, underwriting, and pre-underwriting.

This is because the customer did not clear the background before uploading the policy image, but the customer's image quality cannot be required, and the invalid background information can only be automatically identified and cleared in the post-processing

Images