Vehicle damage detection model training method and device, vehicle damage detection method and device, equipment and medium

Abstract

The invention relates to the field of artificial intelligence classification models. The invention provides a vehicle damage detection model training method and device, a vehicle damage detection method and device, equipment and a medium. The method comprises the following steps: inputting a vehicle loss sample set containing vehicle loss sample images into a vehicle loss detection model for training, extracting loss texture features through the vehicle loss detection model based on InceptionV4 model architecture, and obtaining at least one prediction result; obtaining an identification resultthrough adoption of a GIOU method and a soft-NMS algorithm; determining a first loss value through a GIOU loss algorithm, and determining a second loss value through adoption of a multi-classification cross entropy method; determining a total loss value according to the first loss value and the second loss value; and when the total loss value does not reach the preset convergence condition, iteratively updating the initial parameters of the vehicle loss detection model until the total loss value reaches the preset convergence condition, and recording the converged vehicle loss detection modelas a trained vehicle loss detection model. According to the invention, the vehicle damage type and the vehicle damage area can be quickly identified.

Description

technical field [0001] The invention relates to the field of artificial intelligence classification models, in particular to a vehicle damage detection model training, a vehicle damage detection method, a device, a computer device and a storage medium. Background technique [0002] After a vehicle accident occurs, some parts of the vehicle will leave traces of damage such as damage, scratches, etc. At present, insurance companies generally manually identify images of vehicle damage after traffic accidents taken by vehicle owners or business personnel, that is, to The damage type and damage area of ​​the damaged part of the vehicle in the image are manually identified and judged. In this way, due to the influence of different understanding of standards and insufficient observation experience, the artificially identified damage type and damage area may not match; for example: due to depression and Scratches are difficult to distinguish through visual inspection images, and dam...

AI Technical Summary

Problems solved by technology

[0002] After a vehicle accident occurs, some parts of the vehicle will leave traces of damage such as damage, scratches, etc. At present, insurance companies generally manually identify images of vehicle damage after traffic accidents taken by vehicle owners or business personnel, that is, to The damage type and damage area of ​​the damaged part of the vehicle in the image are manually identified and judged. In this way, due to the influence of different understanding of standards and insufficient observation experience, the artificially identified damage type and damage area may not match; for example: due to depression and Scratches are difficult to distinguish through visual inspection images, and damage assessment personnel can easily determine the type of damage that is sunken as the type of damage caused by scratches. The error in determining the loss caused by the above situation will greatly reduce the accuracy of the damage assessment; While causing the cost loss of the insurance company, it will also reduce the satisfaction of car owners or customers; in addition, the workload of manual loss determination is huge, and the efficiency of loss determination is low. When a certain accuracy of loss determination needs to be met, it will be further improved. workload, reduce work efficiency

Images