Novel Approach of Image Processing using CNN

Abstract

In today’s world where in every aspect of technology and life we see some or the other form of Machine Learning & Artificial Intelligence in automation of task or work force. Machine learning and implementing machine learning in any task is not that difficult when compared to AI. Implementing Artificial Intelligence System and maintenance requires a lot of computing power and expertise. Neural network is such applications of Artificial Intelligence which imitates human brain, and in Neural Network there is branch named as Convolutional neural networks (CNNs), CNN bring best of both worlds machine learning and AI. CNN is easier to implement, and it necessitates less upkeep when compared to conventional AI Convolutional neural networks (CNNs), an intermediate form of adaptive image processing in addition to adaptive filters and conventional feed-forward neural networks, Two-dimensional CNNs are made up of one or more layers of two-dimensional filters with down sampling and/or nonlinear activation capabilities. Invariance of translation and spatially local links are important characteristics of CNNs (receptive fields). The architecture of convolutional networks is described in this study.