Automatic Vehicle Damage Detection Classification framework using Fast and Mask Deep learning

Now a day’s road accidents occurred very frequently in every part of the world. Damage investigation and detection after every accident becomes a very challenging and primary job for vehicle insurance department personnel and vehicle maintenance and rental industries. There is a necessity for some Automatic Vehicle Damage Detection System with the capability of detecting the minor and major damages that are occurred in a vehicle after any type of accident so that an appropriate appraisal can be made and a suitable amount paid to a guilty customer. There is an extensive growth in technology, which leads to the extensive development in Computer System Vision techniques. A tremendous amount of Computer Vision technology has been applied to various real-world life problems. There is a need for an effective automatic vehicle damage system to reduce the vast need for manual manpower required for this damage detection work after the occurrence of an accident. In most of the existing research work, the major problem faced by the researchers is there exist very few datasets to train the system which can be used to classify the damage. And the major challenge is the accuracy of those classifiers to classify the vehicle images into their appropriate damage classes. As the number of vehicles on the roadside goes on increasing, there is a need for automatic systems to ease the post-accident or damage processing goes on increasing. In our research work, we have proposed a Deep Learning-based Vehicle Damage detection assessment algorithm. The algorithm is based on image classification and objection detection methodology capable of automatically determining the actual damage position in a vehicle, degree, and type of damage from the various images received from the guilty user or from any person which will help them to provide a suitable maintenance amount,the calculation done by the insurance company. The proposed work provides a brief idea about the damages occurred during accident and also about the appropriate appraisal.The proposed system focused on fast response to the user, the user don’t have to wait.The user will get the information just by providing the image of the damaged portion of a vehicle after an accident occurred with greater accuracy.