Vehicle Classification Using Deep Feature Fusion and Genetic Algorithms

Vehicle classification is a challenging task in the area of image processing. It involves various vehicles based on their color, model, and make. A distinctive variety belonging to model categories have been developed automobile industry, which has made it necessary establish compact system that can classify within complex group. well-established vehicle applications security, monitoring traffic cameras, route analysis autonomous vehicles, control systems. In this paper, hybrid integration pre-trained Convolutional Neural Network (CNN) an evolutionary feature selection proposed for classification. The performs eight different including sports cars, luxury cars power-house SUVs. used work derived from Stanford car dataset contains almost 196 classes. After performing appropriate data preparation preprocessing steps, learning extraction carried out using VGG16 first learns extracts deep features set input images. These are then taken last fully connected layer VGG16, optimization phase evolution-based nature-inspired Genetic Algorithm (GA). performed numerous SVM kernels where Cubic achieves accuracy 99.7% outperforms other as well excels terns performance compared existing works.