A novel deep learning approach towards estimating the grade of the road ahead of the vehicle through vision is presented in this paper. We have used Inertial Measurement Unit (IMU) pitch values and also Global Positioning System (GPS) altitude values to estimate road grade. Image segmentation has been employed to check if the model performs better when trained specifically on the edges. CNN architecture has been used to predict the grade of the road. We have successfully implemented the model in real time to evaluate the grade resulting in considerable performance. With this paper we aim to give an insight into how a vehicle can change its power distribution if it has the knowledge about the upcoming grade. This would help in improving the fuel economy, ride safety and comfort to quite an extent.