ABSTRACT
Convolutional Neural Networks (CNNs) are considered to be an essential architecture for Neural Network in order to run Deep Learning-based applications. These networks thus achieve a good accuracy in accomplishing task of processing and predicting images, sounds and videos. In Bell Lab, Prof. Yann Lecumn had developed first CNN by late 1990s. Within reasonable training time, Multilayer Perceptron (MLP) produced greater classification results over MNIST datasets. The performance goes on decreasing with increase in training time for large datasets because of parameter size in a model increases exponentially.
A picture as a Landmark can be well recognized with API of Landmark recognition supported in the package of Machine Learning. The images as input parameter is passed to the API, which detects the area it belongs to with the help of geospatial & geographical coordinates. 648This information further helps in generating image metadata by which user shares and creates his experiences. The main purpose of this article is to present an approach for the use of Convolutional Neural Network for historical landmark detection.
