Platforma Remote damage assessment service for machines using a neural network

Main articles:

• Video analytics (terms, applications, technologies)
• Computer vision (machine vision)
• Neural networks (neural networks)

2021: Development of a service for remote assessment of damage in machines

On September 16, 2021, Platforma announced the development of a service that allows you to remotely assess the condition of the body of the car: the scale of damage, including after an accident, the presence of dirt, snow, foreign stickers, etc. The company plans to attract insurance and car sharing companies, as well as automotive marketplaces, to its industrial piloting and subsequent use. Platforma experts estimate the segment of remote valuation of property based on big data at 1.2 billion rubles in 2025.

Remote evaluation of machines works on the basis of neural networks, which are developed by Platforma specialists. Neural networks are trained to recognize car models, injured body parts, types of damage, dirt, snow, etc. During the piloting period, the neural networks processed more than 30 thousand photographs and determined more than 65 thousand elements of the car. As of September 2021, the model stably recognizes 48 classes of parts and 14 classes of damage.

The next stage in the development of the service will be the automatic calculation of the cost of repair, based on the scale of damage. According to Platforma experts, it is this option that is most in demand among potential customers, including banks and insurance companies, so technology is already being developed for the most relevant digitization of repair work.

The company implements an ambitious task - by 2025 to take a high position in the field of services based on Big Data. Platforma specialists use a special amount of data and artificial intelligence technologies to create digital services that can be used in various fields. Remote property assessment using artificial intelligence algorithms is a big step forward in automating the work of insurance and car sharing companies. The service is able not only to increase the efficiency of property valuation processes, but also to significantly improve the quality and speed of customer service, "said VTB Deputy President-Chairman of the Board, Head of the Board of Directors of Platforma Vadim Kulik.

In 2020, the market for remote property valuation services using big data Russia amounted to 700 million rubles, we expect that by 2025 it will increase to 1.2 billion. Abroad, such a practice is already more developed. One of the most striking examples of foreign players can be called British Tractable American and Cape Analytics, which have been operating in the market since 2014 and during this time have collectively attracted more than 90 million dollars investments, "said Platforma CEO. Alexey Kashtanov

Based on the results of testing, the company plans to offer the product in the form of a special application, as well as APIs for integration into the customer's IT systems. Processing a single request takes several seconds. This will significantly save, for example, insurance companies on the departure of their agents - the cost of one trip to Moscow is about 3 thousand rubles, according to expert estimates. Also, the use of neural networks will reduce the potential level of fraud, automate photo processing, etc.

As of September 2021, the service is operating in test mode. An industrial version for wide rolling will be available in the first quarter of 2022.