2020 EE-5183-Financial-Technology Final Project
Our research focuses on the area of insurance and we aim to make the claiming process more simpler and shorten the time to get the claims after filing an car insurance claim. We build a car damage detection model, which is composed of 4 submodels, to detect whether the photo is a car, whether the car is damaged, which part is damaged, and the severity of damage. Our proposed method combines deep learning, instance segmentation, and transfer learning techniques for features extraction and damage identification, also with a promising attempt in classifying car damages into a few different classes. Along the way, the main focus was on the influence of certain hyper-parameters and on seeking theoretically founded ways to adapt them, all with the objective of progressing to satisfactory results as fast as possible. This research open doors for future collaborations on image recognition projects in general and for the car insurance field in particular.
Our proposed method combines deep learning, instance segmentation, and transfer learning techniques for features extraction and damage identification, which is composed of four models. Our objective is to automatically detect damages in cars, locate them, classify their severity levels, and visualize them by contouring their exact locations.
[1] “What is auto insurance? — iii,” https://www.iii.org/article/what-auto-insurance, (Accessed on 01/13/2021).
[2] “What to expect when filing a car insurance claim — allstate,” https://www.allstate.com/tr/car-insurance/howto-file-auto-insurance-claim.aspx, (Accessed on 01/13/2021).
[3] Arun Mohan and Sumathi Poobal, “Crack detection using image processing: A critical review and analysis,” Alexandria Engineering Journal, vol. 57, no. 2, pp. 787–798, 2018.
[4] Srimal Jayawardena et al., “Image based automatic vehicle damage detection,” 2013.
[5] Kalpesh Patil, Mandar Kulkarni, Anand Sriraman, and Shirish Karande, “Deep learning based car damage classification,” in 2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA). IEEE, 2017, pp. 50–54.
[6] Pei Li, Bingyu Shen, and Weishan Dong, “An antifraud system for car insurance claim based on visual evidence,” arXiv preprint arXiv:1804.11207, 2018.
[7] Soumalya Sarkar, Kishore K Reddy, Michael Giering, and Mark R Gurvich, “Deep learning for structural health monitoring: A damage characterization application,” in Annual conference of the prognostics and health management society, 2016, pp. 176–182.
[8] Young-Jin Cha, Wooram Choi, and Oral Buy¨ uk¨ ozt ¨ urk, ¨ “Deep learning-based crack damage detection using convolutional neural networks,” Computer-Aided Civil and Infrastructure Engineering, vol. 32, no. 5, pp. 361–378, 2017.
[9] Ranjodh Singh, Meghna P Ayyar, Tata Sri Pavan, Sandeep Gosain, and Rajiv Ratn Shah, “Automating car insurance claims using deep learning techniques,” in 2019 IEEE Fifth International Conference on Multimedia Big Data (BigMM). IEEE, 2019, pp. 199–207.
[10] “car-damage-dataset.tar.gz - google drive,” https://drive.google.com/file/d/1IYx9kCFgCWUJgudc8Uxn59CkExRei431/view,(Accessed on 01/13/2021).