Maximum precision and speed
A unique artificial intelligence model
E-IA analyzes the vehicle and generates a report with maximum precision and speed.
The performance of an HLP approach is subjective and characterized by variability
E-IA is trained and continuously updated
E-IA has been trained to “see,” highlight, and count hail damage on the surface of cars, taking into account the parameters and conditions set during the training phase.
E-IA is trained on continuously updated databases: it is able to consistently provide objective and updated appraisals with performance superior to that of humans.
Currently, 20-40% of appraisals are reopened due to:
- Incorrect numerical evaluation of the damage (with errors ranging from 50% to 80%)
- Incorrect evaluation of the type of intervention to be performed on the bodywork (Repainting/Replacement Operations).
E-IA is a unique artificial intelligence model in the world. E-IA analyzes the vehicle and generates a report with maximum precision and speed.
EIA’s response is certified to ISO standards.
E-IA is regularly calibrated and results are provided, as per standard, through specific confidence intervals.
Countin
Performs a precise and detailed count of impacts on the vehicle's bodywork, providing an accurate estimate of the damage sustained.
Identification
Identifies the type of intervention required for repair, including total PDR, a combination of PDR and traditional interventions, or pure traditional.
Identification
Detects and catalogs the bodywork elements that have been damaged and require replacement or specific repair.
Reporting
Reports any additional damages not related to hail, helping to provide a comprehensive assessment of the vehicle.
Final Triage
Determines whether the vehicle is repairable or should be considered a total loss, based on a thorough analysis of the detected damages.
E-IA Training
For accurate identification of all impacts, the metamerism-based model underlying WeGrele technology has been trained to analyze images from various angles and positions in order to faithfully reproduce a "physical" evaluation.
The model has been trained to identify hail impacts on vehicles in various conditions, including:
- The color of the vehicle.
- Weather conditions.
- The condition of the vehicle (more or less clean). The surrounding environment (interior or exterior). The reflections on the bodywork.
- Trained on devices with different types of optical HW and SW technologies, to be able to recognize any image manipulations. Bias management through dedicated calibration for each specific area of the bodywork extracted.
- Real data validated by a professional technician.
Technical Features
The Mask RCNN-FPN model
The model uses an architecture called “Mask RCNN-FPN” which excels in object detection and instance segmentation, leveraging a multi-scale feature pyramid. This approach ensures robust performance, adapting to variations in object sizes and offering compatibility with various backbone networks, achieving state-of-the-art results on benchmark datasets.
RCNN = Region-based Convolutional Neural Network; FPN = Feature Pyramid Network.
A robust Dataset is employed (approximately 10,000 instances) using various Data Augmentation techniques: The dataset is constantly expanding and will be used to retrain the network to improve its performance.
This photographic approach, both for the training phase and for the user during use, guarantees a better characterization of the vehicle, allowing detailed information to be obtained on the status of the bodywork and vehicle components.
The Mask RCNN-FPN model
EIA uses advanced training techniques
To optimize the generalization capability and performance of the artificial intelligence model, we have used the Data Augmentation technique.
Data Augmentation is used to increase the amount of data available for training an artificial intelligence model. This technique involves creating synthetic data from existing data, such as rotating, transforming, and modifying images.
EIA continues to grow...
Update: Regular update of the training database.
Upgrade: New AI technologies will be regularly implemented and updated.
Regular updates and upgrades will involve the use of new features:
- Dimensional identification of individual damages to promptly assess the repairability of various sections of the vehicle online
- Implementation of capabilities dedicated to specific vehicles or types of damage; Online assessment of the cost-benefit ratio based on the relationship between EIA’s damage assessment and the value of the car (year, model, market value of the vehicle).
E-IA plus
Time saving
- Extremely fast damage analysis
- Reduction of waiting times
- Speeding up the Financial planning of companies
Cost reduction
- Fraud reductions
- Reduction of staff to mobilize
- Correct positioning in the repair process
Accuracy
- Objective evaluation through certified algorithm
Easy to use
- Completely digital and easy to integrate
- Reduction of waiting times
- Ideal for all claims in the most difficult to reach areas
A new, very easy image acquisition process
Through a few simple steps, you will be able to start the claims reporting procedure directly from your mobile phone. You will need to enter the documents and take photos of the vehicle through our App, which will guide you step by step, even indicating how many and which photos are necessary for an accurate damage assessment. Once the photos are uploaded, our advanced E.I.A. technology analyzes the damage and provides a detailed and precise report.
