Review on the Methodologies for Image Segmentation Based on CNN

Abstract

In computer vision, image distribution is the process of dividing the digital image into several categories. The main reason for the division is that, the image should be divided into regions for better analysis. Today, the distribution method is used for applications such as image classification, facial recognition, object, and video analysis. Several computer vision functions require some knowledge base using which the visual classification is carried out in order to identify the contents. Recent surveys on image distribution techniques often focus on deep learning techniques, to accurately identify real-world objects within each image. Techniques for segmenting several types of objects using numerous algorithms have been presented in the literature, but most are based on a complex strategies and little work has been reported on systematic interpretation. To provide a comprehensive and lucid approach for segmentation of the objects, deep learning approaches are the most appropriate. These methodologies will purposefully study the characteristics of visual aids and help in better identification of the components within the objects more comprehensively. Image Processing can be done using Deep learning design in a Convolutional Neural Network (CNN). These models are distinctly trained and implemented in engrossed processing units like GPUs (Graphical Processing Units), so that the execution time can be better optimized. In the article, other variations of CNN will also be articulated for more elegant object detection.