In today's world, a GPS based navigation system is inevitable. Though GPS provides a better navigation system, it compromises one's location privacy. Unintentional broadcast of one's location may open the gateway for malicious 3rd party applications that may not operate on a legal basis. To overcome privacy related issues, this paper proposes a method of navigation without the use of GPS. This can be achieved by combining Convolutional Neural Network(CNN) along with Dijkstra’s Algorithm. In this work, initially, the starting point of the user is identified by capturing the image of the nearest building, thereafter using the CNN model to process the image and identify the location. The dataset of more than 6,000 images was built considering various circumstances for training and testing of models. This dataset was fed to different deep learning models and CNN performed the best. The shortest path between the locations is calculated using a modified Dijkstra's algorithm. It uses priority queue to reduce the time complexity of traditional Dijkstra’s algorithm. This paper will help to navigate through different college campuses/universities and other big premises without using GPS.