• Mansi Ramdas Satpute
• Sahil Kadadekar
• Dr. Rupesh C. Jaiswal

In this paper, we have designed and used a car injury severance pipeline, which can be used by insurance companies to automate car insurance claims. Recent developments in Computer vision are largely due to the adoption of fast, scalable, and end-to-end training networks of Convolutional Neural Networks (CNNs) making it technically possible to detect motor vehicle damage through deep transforming networks. We collected and commented on various online sources using a web browser containing various types of vehicle damage. Due to the small size of our database, we have used pre-trained large models and various data sets to avoid overcrowding and to learn common features.Using pre-trained CNN models ImageNet data set and advanced processing techniques to improve our system's performance, we obtained 96.39% accuracy, which is much better than the results obtained in the past in the same test set. In addition to locating the damage site, we used the YOLO modern scanner and obtained a high score of 77.78% on the captured test set, indicating that the model was able to successfully detect various vehicle damage. In addition, we have also developed a solid pipeline for identifying damage to vehicles by integrating segmentation and acquisition activities. All in all these results paved the way for further research on this problem area and we believe that the collection of various databases will be sufficient to use the automated vehicle diagnostic system soon.

machine learning, CNN ,

"Machine Learning Based Car Damage Identification", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.9, Issue 10, page no.b684-b690, October-2022, Available :http://www.jetir.org/papers/JETIR2210195.pdf

"Machine Learning Based Car Damage Identification", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.9, Issue 10, page no. ppb684-b690, October-2022, Available at : http://www.jetir.org/papers/JETIR2210195.pdf