Challenges
- Identify and localize small dents caused by hailstorm on vehicles for an Insurance application
- All small dents had to be detected whereas an actual gap like between the doors or curves as part of panel design should not be misclassified
- Detect small hailstorm dents which could be only a few pixels in the image, while noise should not be classified as false positives
Solutions
- Develop and deploy an advanced NN model to classify dents caused by hailstorm
- Choose a model whose feature extraction capability does not let small features vanish as we go to deeper layers
- At the same time, it could improve speed and accuracy for small dent detection on low resolution images
- Retrain the model whenever there is any misclassification or when a new dent shape is detected
Benefits
- Accurate and automatic detection of dents
- Since there is no manual intervention, this solution enables improved and high-speed insurance settlements without any frauds during hailstorm season
