Novel Framework for the Improvement of Object Detection Accuracy of Smart Surveillance Camera Visuals Using Modified Convolutional Neural Network Technique Compared with Global Color Histogram

Abstract

The objective of the proposed work is to evaluate the accuracy of Global Color Histogram (GCH) and Modified Convolutional Neural Networks Techniques (MCNNT) by changing the hierarchical data processing for Smart Surveillance System. Methods and materials: Our novel framework with MCNNT uses the data in hierarchical models of data processing, which is constructed with fully connected layers and the neurons were connected to every other node to handle complex object detection tasks and evaluated using the sample size of 20 per group with model image dataset. Result: The recorded mean accuracy of MCNNT (95.28%) is better than GCH (80.48%). There is a statistical significant difference in accuracy for two algorithms performed by independent sample t-test (P<0.05) for the Confidence Interval (CI) 95%. Conclusion: MCNNT appears to have better accuracy than GCH for object detection of Smart Surveillance System.