Multiple Object Detection Using Mask R-CNN in Deep Neural Network

  • Gowri Shankar V, Arun kumar S, Dr. Balika J Chelliah

Abstract

In multi-mark order, each example can be related with a lot of class names. At the point when the quantity of names develops to the hundreds or even thousands, existing multi name characterization strategies frequently become computationally wasteful. Lately, various cures have been proposed. In any case, they depend either on straightforward measurement decrease systems or include costly advancement issues. Mask R-CNN has exhibited promising execution in single name picture order undertakings. Be that as it may, how Mask R-CNN best adapts to multi-mark pictures despite everything stays an open issue, for the most part because of the complex basic item designs and lacking multi-name preparing pictures. Multi-name characterization is a more troublesome assignment than single-name grouping in light of the fact that both the information pictures and yield mark spaces are progressively unpredictable. Multi –labeling becomes progressively intense when we attempt to name with increasingly jumbled picture, for example, in untamed life pictures where there is more than one creature in one picture alongside the trees and scenes, we will attempt to improve the precision of a Mask R-CNN model so that it can multi-name the pictures all the more precisely and absolutely.

Keywords: Region Proposal Network, Mask R-CNN.

Published
2020-04-28
How to Cite
Gowri Shankar V, Arun kumar S, Dr. Balika J Chelliah. (2020). Multiple Object Detection Using Mask R-CNN in Deep Neural Network. International Journal of Advanced Science and Technology, 29(05), 3677 - 3686. Retrieved from http://sersc.org/journals/index.php/IJAST/article/view/12260