Comparative Analysis of Convolutional Neural Network in Object Detection

Abstract

Deep learning has proven to be a vital technique over the last few decades based on its potential to manage a huge amount of data. It is an artificial intelligence function that mimics the functioning of the human brain in accessing information and generating patterns to be used in problem solving. Using its methodologies, complex computer vision tasks like object detection including facial recognition and image classification can achieve cutting-edge results. With the help of deep learning object detection algorithms, we can instantly recognize and localize objects of interest once we look at pictures and videos with a meaningful result. The primary objective of object detection is to mimic this intellectual ability or intelligence using a computer. Object detection can be accomplished using a variety of deep learning techniques like YOLO and R-CNN including FASTER R-CNN and FAST R-CNN, and all these techniques use convolutional neural networks (CNN). We implemented these deep learning techniques to object detection in this paper as well as analyzed the results to choose which one performs better with the highest accuracy and we found that YOLO works better among four of them. However, the performance of these algorithms varies according to the scenarios in which they are used.