A SURVEY OF ARTIFICIAL NEURAL NETWORKS AND SEMANTIC SEGMENTATION

Abstract

An Artificial Neural Network (ANN) is an information processing paradigm that is inspired by the way biological nervous systems. Tasks that requires human like intelligence or trends that are too complex to be noticed by either humans or other algorithm techniques can be solved by the remarkable ability of neural networks to derive meaning from complicated or imprecise data, and this can be used to extract patterns from the given data. A trained neural network can be thought of as an expert in the category of information it has been given to analyses. This expert can then be used to provide meaningful information from an image or video Segmentation is a partition of an image into several coherent parts in terms of low-level cues such as colour, texture and smoothness of boundary. Semantic segmentation tries to rupture the image into semantically purposeful parts and classify each part into one of the pre-determined classes. This process achieves the segmentation goal by classifying the image in a pixel level rather than the entire image. Semantic segmentation is a complex task requiring knowledge of support relationships and contextual information, as well as visual appearance. Early methods that relied on low level vision cues have fast been superseded by popular machine learning algorithms. Deep convolution neural network can be employed to perform such machine learning tasks semantic segmentation. The challenge in performing semantic segmentation is that the scene may contain foreign objects as well as scenes often vary significantly in pose and appearance. This survey gives an overview over different techniques used for pixel-level semantic segmentation and hierarchy of artificial neural networks. The very recent approaches with convolution neural networks are mentioned which is widely being used now and the taxonomy of segmentation algorithms is given.