Enhance your patent search with AI. Try the FREE AI-powered tool

Accenture is patenting AI that can tell how cars get damaged

Pexels/Aleksandr Neplokhov
July 9, 2021

Inventors from Accenture have developed artificial intelligence that uses images of a car to predict causes of damage, along with costs of repair.

The Ireland-based company’s patent application uses vehicles to illustrate the invention’s functionality, but its potential use can be extended to damage detection and cost estimation for buildings, materials, and other types of objects.

Accenture mentions how the technology leverages big data. In the case of the auto insurance industry, which is worth at least $739 billion globally, the technology could assist professional assessors and expedite the process for claimants.

Accenture is patenting AI that can tell how cars get damaged

The invention would estimate damages using images or videos of an automobile, captured through a mobile device application, in color or in monochrome. The system may incorporate a method to convert RGB inputs to black and white, as needed. Accenture says their machine learning (ML) model was shown to be more accurate when using monochrome data.

The AI could hypothesize how a car got damaged from a list of causes including collisions, hailstorms, and natural perils. This assessment may come with a probability percentage, along with the specific pixels of the image that informed the ML algorithm’s hypothesis.

Additionally, the system may label the parts of the vehicle that can be repaired or must be replaced. The AI may also be trained to provide a cost estimate for its suggested fixes.

To train an AI system for damage assessment, the inventors propose using images of objects, both damaged and undamaged. These training data would come with labels for corresponding damage causes. The supervised ML aims to cover various parts of various vehicles, for example, along with various types of damage from various sources.

Inventors claim the method produces a ML cause prediction model, with sub-models trained for different damage causes for different angles of a vehicle. This ensemble approach increases accuracy by enabling specialized ML models to cover each other’s weaknesses.

The patent application states that while convolutional neural networks are already employed for image analysis tasks, they produce assessments that are often incomprehensible. Inventors attribute this to a black box approach born out of data privacy and other regulatory issues.

Researchers have only recently explored breaking away from the idea that the most accurate AIs are inherently uninterpretable and complicated, even for the people who designed them. Not only does Accenture’s invention enable automatic damage assessment, it may be among the first to do so in a way that people can easily understand.

The featured patent application, “Explainable Artificial Intelligence (AI) Based Image Analytic, Automatic Damage Detection and Estimation System”, was filed with the USPTO on December 16, 2019 and published thereafter on June 17, 2021. The listed applicant is Accenture Global Solutions. The listed inventors are Indrajit Kar, Vishal D. Pandey, Mohammed C. Salman, and Ankit Vashishta.

Related Stories

Subscribe to our newsletter

  • Questions? Check our privacy policy.
  • This field is for validation purposes and should be left unchanged.

Patent and IP updates straight to your inbox

Sign up now to receive monthly patent news, analysis and free insight reports.

We don’t spam, we promise.

Disclaimer: 

1. Parola Analytics and Avontis are distinct entities and operate independently. Any references to Avontis or its services do not constitute a legal partnership. 

2. Parola Analytics does not provide legal services. Our services are limited to research and technical analysis. Any information provided by Parola Analytics should not be construed as legal advice.