K Prashanth

Automotives & Insurance Technique: Deep Learning (Transfer Learning, Mask R Title: Automated car exterior damage assessment

The primary objective of this project is to develop and validate machine learning models that can automatically detect and assess the severity of car damage using images. By leveraging transfer learning and fine tuning of foundation models which run on convolutional neural networks (CNNs), the model aims to accurately classify the type and severity of damage, facilitating quicker and more consistent damage assessments.

This project focuses on applying CNNs enhanced by transfer learning techniques to classify car damage into multiple categories based on severity and type. The scope includes the development of a robust model trained on a curated dataset of car images with varying damage types, the evaluation of model performance through rigorous metrics, and a discussion on the feasibility of deployment in real-world scenarios.