AI AUTOMATIC DAMAGE DETECTION

Vehicle types: Systems typically handle passenger cars (sedan, hatchback, SUV), some suppliers also work with vans, pickup trucks and trucks.
 

Detected damages: The AI (artificial intelligence) typically detects scratches and dents on the surface of the vehicles. In addition, some systems can also detect damages on alloys and wheels, broken windscreens, differences in color of the parts, and gap dimensions. Some systems can analyze the underbody of the car. An analysis of the vehicle interior we have seen nowhere yet, but some suppliers are working on this.
 

Generic vs. make/model: A key differentiator is whether the AI is based on specific make/model information or is generic. In the first case, specific information is required to be uploaded to the system before, while in the second case the AI works on a bodystyle-level, which means no specific make/model data is required.
 

AI quality: While one should expect that all suppliers of automatic vehicle inspections/ damage detection use AI (artificial intelligence), one should check carefully what is behind it. The key criteria is the quality of damage detection, which differs significantly. In some cases, systems rely still on manual assessments from remote assessors, while others are fully automated.
​

Processing time: Processing time varies between 1 minute per inspection and up to 15 minutes. This is mostly dricen by the power of the AI and the need for manual corrections. Important driver is also the number of pictures taken: Some systems require several hundred pictures, which slows processting time down.
 

Cost calculation: Basically two ways exist to estimate the repair costs: Either the system is integrated with third party data providers like DAT, gtMotive,... This enables exact cost calculations based on standard labour values, parts prices etc. Downside of this is an additional cost for the data provider. Alternatively, systems can calculate costs based on an internal database. This is usually less detailed, but can be sufficient for some use cases.
 

Before/after comparison: Some systems offer the feature to compare pictures of a vehicle at different points in time. E.g. pictures of the car when leaving in the morning, and when returning in the evening. The AI automatically detects additional damages that happened in the meantime.
 

Further detections: Some systems offer to also automatically detect more information like VIN, number plate, mileage from the odometer or information from vehicle papers.
 

System integration: Differention criteria is whether the supplier is able to provide an own App or user interface and/or is able to integrate into existing systems of the customers, like vehicle inspection systems, dealer management systems, pricing tools, used car management systems or trading platforms.
 

Interior: Often requested is the functionality to automatically detect damages of the interior of a vehicle. So far, no systems are available with this feature, but some suppliers are working on it.