An Intelligent Traffic Management of Vehicles using Deep Learning Approach in Smart Cities

Smart traffic management of automobiles has been one of the utmost sought-after issues in the academic community due to the ever-increasing pace of urbanization and civilization over the last several centuries. Vehicle dissection, traffic density appraisal, and vehicle tracing are the three main components of smart traffic management. This problem becomes more difficult to resolve when there are occlusions, background clutter, and fluctuations in traffic density. In light of this need, we explore a deeper learning approach based on R-CNN s for faster vehicle segmentation in this study. There is adaptive backdrop modeling in the computational framework. It solves problems with lighting and shadows as well. The use of topological active net deformable models allows for segmentation that is more precise. The deformations mentioned can be accomplished with the aid of topological and stretched topological active nets. Minimizing energy allows for mesh deformation. Using a tweaked version of the stretched topological active net improves the segmentation accuracy to 98.3%. The investigational findings show that this framework is better than other methods.