Convolutional Neural Networks (CNNs) for Image Recognition and Detection

International Journal of Research and Scientific Innovation (IJRSI) | Volume VI, Issue I, January 2019 | ISSN 2321–2705

Convolutional Neural Networks (CNNs) for Image Recognition and Detection

Aparna Mohan

Department of Computer Science and Engineering, Jawaharlal College of Engineering and Technology, Mangalam, Lakkidi, Ottappalam, Palakkad, Kerala, India

Abstract – Nowadays, face recognition is widely uses in many security based applications. Even mobile phones and other such gadgets consider face as one of the most secure biometric application. Deep learning based models are used for face recognition. Deep features are obtained by using several convolutional and pooling layers to extract features from input images.

Keywords – Convolutional neural network (CNN), face recognition, LBP, face detection, texture classification

I. INTRODUCTION

Convolutional neural network is widely used in pattern and image recognition. Because they have number of advantages compared to other techniques. Neurons exchange messages between each other. A neural network is a system of interconnected artificial “neurons”. Each connection have its own numeric weights, tuned during the training process. In each of the network there are multiple layers of feature detecting “neurons”. Each of these layers consists of many neurons that respond to different combination of input. Human face consists of complex multidimensional meaningful visual stimuli. Therefore developing a computational model for human face recognition is a difficult one. Hence convolutional neural networks (CNNs) have been established as a powerfull class of neural network models among all others for image recognition problems.

In today’s era images and videos have become ubiquitous over the internet. This leads to the development of algorithms that analyze the semantic content of various applications. For the better understanding of image content, statement of the art, results on image recognition, segmentation, detection and retrieval the convolutional neural network have been considered as an effective class among other neural network models.

Read More