Road Feature extraction using CNN and GRU in edge-enabled vehicular cloud networks

Efficient and accurate road feature recognition has a vital part in securing road safety and enhancing the driving experience. Poor road conditions, such as potholes, uneven surfaces, and lack of proper road markings, can lead to a higher risk of accidents, including vehicle damage and personal injury. In this study, we present a hybrid model (CNN+RNN) for the improved categorization of road characteristics. Specifically, we use RNN to take on GRU and Convolutional Neural Networks (CNN) based on MobileNetV2 architecture.MobileNetV2's lightweight architecture ensures real-time processing and model portability, making it a valuable asset for in-vehicle applications. Existing LSTMs are computationally expensive, especially when dealing with long sequences. CNN-MobileNetV2-GRU model provides enhanced capabilities for road feature recognition. We evaluate our model on extensive road feature datasets and demonstrate its superior performance in comparison to the existing CNN-LSTM model. This is particularly beneficial for detecting dynamic road conditions and speed bump occurrences, as it considers the evolution of features over time. By integrating these two components, our model offers a holistic solution for road feature classification. We here process the data collected from edge to the cloud, which requires a lot of space, efficiency, and fastness for these requirements this hybrid algorithm is best. The CNN-MobileNetV2-GRU model not only exhibits higher accuracy but also operates with reduced computational resources, making it suitable for uses in the real world, such as driverless cars and advanced driver assistance systems (ADAS). Our research suggests that this strategy can greatly enhance both road safety and the driving experience in general. fastness for these requirements this hybrid algorithm is best. The CNN-MobileNetV2-GRU model not only exhibits higher accuracy but also operates with reduced computational resources, making it suitable for uses in the real world, such as driverless cars and advanced driver assistance systems (ADAS). Our research suggests that this strategy can greatly enhance both road safety and the driving experience in general.