The paper focuses on the implementation of an optimized, cost effective parking solution for smart cities where car will be parked and guided on place or a block where it best suits and the parking area is fully utilized with minimum quanta of fragmentation or left over space. This paper includes utilizing algorithm of operating system, concept of image processing and image analysis through convolutional neural network and a portal using a framework with database connectivity. The Proper dataflow is being descripted from the car image acquisition and its processing a basic predictions and the allotment of car in a particular block of a parking system. This approach would be beneficial for the maximum utilization of that area also the specific usage in hospitals, malls etc would not face any problem. The main key role is to include deep learning in the model fitting for training and testing the data set containing images of cars and there categories where they belongs. The methodology used for allotment of cars is the fitting algorithm used in operating systems and refining, modifying the algorithm according to the suitability of the dynamic car allotment in real time. The database developed will comprise which block is empty or occupied and left over space will be there for the proper guidance of cars and for the calculation shown in mathematical approach.