Russian IT-company Platforma has developed a service capable of assessing the level of car body damage from photographs, Inc.Russia reports.
The service identifies damage with the help of neural networks. They can recognize the models of cars, damaged parts of the body and contamination on it. Neural networks were trained on 30 thousand photos and 65 thousand elements of the car. As a result, the platform can steadily recognize 48 classes of cars and 14 classes of damage. In the future the developers intend to teach the neural networks to calculate the cost of repairs.
It is assumed that the service will run in the form of an application and API for integration into the customer's IT-systems. Apart from drivers, the system can be used by insurance companies, thus they will have to involve insurance agents less frequently.
At the moment the service is operating in the test mode, its full launch is planned for the first quarter of 2022. For the industrial piloting and subsequent use of the service Platforma plans to attract insurance, car sharing companies and car marketplaces.
Made in Russia // Made in Russia
Author: Karina Kamalova