Car Guard: AI for damage inspection

Car Guard is the intelligent solution developed by Interlogica to make the process of identifying damage on rental vehicles more efficient and reliable.
The system is based on Image Forensics techniques and AI models, with the aim of automating damage assessment, reducing the risk of fraud and speeding up refund operations.

How Car Guard works

1. Image intake and integrity check
   The flow starts when a customer returns a rental car and sends a photo of the damage.
   The first phase is an automatic analysis of the image to verify its:

   • quality (sharpness, correct framing);
   • originality, using Image Forensics techniques to spot tampered or duplicated images.

   This step filters out invalid content and prevents fraud attempts from the very first input.

2. Damage detection
   Once the image has been validated, an AI system locates the position of the damage and highlights it visually with a bounding box.
   This phase is not a simple binary classification (damage or no damage): it returns the precise location of the damaged area.

3. Automatic classification of damage severity
   Next, a generative AI model analyses the highlighted portion of the image and provides an automatic classification of the severity of the damage.
   Based on the analysis, the system suggests a recommended action, for example:

   • No action needed
   • Repair
   • Replacement of the component

   This automation standardises the process and reduces the variability of manual assessments.

Initial focus: windscreen damage

The first operational implementation of Car Guard focuses on windscreen damage, a very frequent case with a high risk of disputes.
The project was designed to act in a targeted way, starting from a well-defined area and then extending to other types of damage on different parts of the vehicle.

Benefits and opportunities

Car Guard introduces a new approach to damage management in vehicle rental, with tangible advantages:

• Fraud elimination thanks to the automatic check of images;
• Cost reduction through faster and more reliable assessments;
• An expandable model, which in the future could cover damage to bodywork, interiors and other components;
• Applicability beyond rental, for example in insurance and after-sales services.

Car Guard is a system designed to combine AI, visual analysis and automated processes, offering a concrete solution for optimising damage management in the automotive sector.