A video-based self-service damage assessment method and system for vehicle exterior parts

Abstract

The invention discloses a method and system for self-service damage assessment and collection of vehicle appearance parts based on video, including: checking whether the posture of the user holding the mobile phone and the ambient light meet the preset conditions; if the preset conditions are met, the video frame is obtained through the mobile phone, and Real-time judgment of the category of the shooting status of the current frame; if the category of the current frame is "vehicle orientation side is easy to identify the orientation", start recording through the mobile phone and prompt "close to the damaged part"; in the process of approaching the damaged part, if the category of the current frame If it is "the surrounding environment of the vehicle part is easy to identify the part", it will prompt "close to the damage details"; in the process of approaching the damage details, if the category of the current frame is "the surrounding environment of the component damage is easy to identify the damage details", it will prompt "keep still" ;Start the countdown and perform frame difference algorithm on the video frame. When the trigger conditions of time and video frame difference are met, the video recording is successful and the recording ends.

Description

technical field [0001] The invention relates to the technical field of vehicle loss assessment, in particular to a method and system for self-service assessment and collection of vehicle appearance components based on video. Background technique [0002] Vehicle appearance parts: usually refers to the appearance components of the vehicle, such as bumper assembly, fender assembly, door assembly, tire hub, etc., excluding the engine, water tank and other accessories that can be seen after opening the front cover. Damage determination: According to the principle of automobile structure, through scientific and systematic professional inspection, testing and survey means, comprehensively analyze the scene of automobile collision and accident, and use vehicle damage estimation data and maintenance data to conduct scientific and systematic evaluation of vehicle collision repair loss pricing. [0003] Since 2012, deep learning has made great progress in the field of image recogniti...

AI Technical Summary

Problems solved by technology

[0010] 1. The cost of on-site investigation by staff is high. In reality, some traffic accident cases have minor vehicle damage, such as reversing and scratching the wall. The conclusion of these cases is simple to determine the damage, but the unified on-site inspection method consumes a lot of staff energy, and also increases the operation of insurance companies. cost;

[0011] 2. The shooting requirements of loss assessment technology are high: without special training, it is difficult to shoot a picture sequence that meets the requirements of industry standards and the expectations of the intelligent loss assessment system. Pictures or videos of natural scenes are not convenient for intelligent algorithms at the current technical level Identify developmental and detrimental data normative storage and information mining

[0012] 3. The existing intelligent damage assessment system usually uploads pictures or videos to the server by the reporter’s mobile phone, and then uses artificial or intelligent algorithms to identify the damage of the pictures, thereby completing the damage assessment process; There is no basic understanding of the process, and it usually happens that the pictures or videos taken are too far away, so that the human eye cannot see the details of the damage, which makes the intelligent damage assessment system fail or the manual identification fails, so the damage assessment specialist will usually contact the reporter for a second time Collect supplementary details, reduce customer experience, and extend claim settlement cycle

[0013] 4. In addition, most patents such as [CN110287768A] do not explain how to collect pictures or videos. In the long run, the absence of scene constraints is not conducive to the development of intelligent loss assessment system algorithms

Images