Author(s) :
Neela K, Devianugraga M. S, Anurega T R
Conference Name :
7th International Conference on Recent Innovations in Computer and Communication (ICRICC 23)
Abstract :
One of the most significant issues in modern road safety and intelligent transportation systems is the automation of vehicle detection and identification. Many challenges have been solved in the advancement of image processing, pattern recognition, and deep learning technology in order to accomplish this goal. Vehicle Type Classification is a difficult task since the dataset has a large class imbalance, and several view points for different cars can be identical. The proposed framework employs a shallow Convolutional Neural Networks (CNN) architecture to prevent overfitting and ensure that the correct features are learned, and we use an augmentation technique to produce synthetic images using the image data generation model in Keras due to class imbalance. The shallow CNN is used to extract features from the generated images, and then Softmax activation is used to classify them. Finally, the proposed system will achieves the classification of vehicle type i.e. classify the different car models with efficiently by novel methodology. The findings of the experiments demonstrate that shallow CNN can do well in real world situations.
No. of Downloads :
1