Robust Deep Learning Model for Real-Time Vehicle Detection in IoV Applications

Crucially important for traffic management, autonomous navigation, and road safety, vehicle recognition is the pillar of intelligent transportation systems (ITS). This work offers a convolutional neural network (CNN)-based architecture meant to solve the difficulties in vehicle recognition under various environmental settings. Using cutting-edge designs including ResNet and YOLOv4, the model combines regularizing methods, data preparation, and transfer learning to improve performance.With data augmentation and hyperparameter tweaking, the approach consists in strong training procedures that significantly increase both training and testing accuracy. With a test accuracy of 95% and a clear drop in test loss, experimental data show the great dependability of the model. Effective generalizing shown by validation loss patterns helps to minimize overfitting and guarantees applicability to unprocessed data.From vehicle identification to classification, these results emphasize the promise of CNN-based models for practical ITS uses. While suggesting future improvements, such including multi-modal data and optimizing for edge devices, the paper ends with addressing constraints including deployment scalability and environmental robustness. This study helps ITS models to be more accurate and efficient, therefore opening the path for more effective vehicle recognition systems.