Adaptive Vehicle Recognition: Leveraging Channel Attention in InceptionV3 for Traffic Optimization

As traffic systems grow in complexity and vehicle numbers increase, traditional management methods struggle to meet real-time demands. Intelligent Transportation Systems (ITS) offer a solution by integrating information and communication technologies for efficient traffic flow monitoring. This paper presents an advanced vehicle classification model, based on the InceptionV3 architecture and enhanced by a channel attention mechanism, to address challenges in fine-grained vehicle recognition under varying conditions such as diverse lighting, multi-angle perspectives, and occlusions. Tested on a comprehensive vehicle dataset, the model achieved 97% classification accuracy, significantly surpassing baseline models. Key contributions include a refined vehicle classification model for ITS, enhanced strategies for handling occlusions and multi-angle scenarios, and robust experimental validation across classification metrics, providing a solid foundation for ITS-driven traffic optimization and safety management.