An Efficient Deep Learning Based Vehicle Classification System for Indian Vehicle

Image classification is heavily used in many technologies. Applications like Face identification, Video surveillance are some practices of image classification. Vehicle Classification is one of the critical domains of Image Classification. This research aims to find efficient methods for Vehicle classification that help solve real-world problems like security, surveillance, and traffic jams. The dataset was used from Kaggle Website for training the models. We have augmented the data to achieve better accuracy of models. In this research, we have focused on Deep Neural Networks which provide more accuracy. We will be using a Simple 4-layer Convolutional Neural Network and Pre-built Models like AlexNet, VGGNet, and DenseNet. After processing our augmented data, we fed it to the CNN model and compared the different models’ results. It was found that DenseNet is having highest accuracy (87%) on the vehicle dataset. Furthermore, this method can be useful for any type of image classification task. Training the model for different weather conditions such as winter, monsoon, etc. is possible. The models can be helpful in many sectors for example in Tumor detection in MRI reports in healthcare or for identifying different cracks and damages of building in the civil sector.