The Convolution Neural Networks have several different filters (also known as kernels) that consist of parameters that can be trained. Convolution Neural Network convolves any given image by the user spatially. It detects characteristics such as shapes, edges and corners. The large number of filters are highly efficient and effective. They learn to extract spatial features from any given image that are certainly based on the learned weights by back propagation approach. Layers of filters when stacked can be used to detect the spatial shapes which have a high level of complexity from the spatial features at every subse-quent level. Therefore, the layers of filters can successfully extract the charac-teristics of a given image by considering the edges and vertices into an abstract-ed representation of high quality. Patterns in pixel values are read and extracted from the given input images in Dense Networks.