Precision Analysis of Salient Object Detection in Moving Video Using Region Based Convolutional Neural Network Compared Over Optical Character Recognition

Abstract

To detect the number plate of the vehicles which violates the traffic rules using optical character recognition compared over region based convolutional networks. Materials and methods: In this work, number plates are identified by optical character recognition with the sample size of 22 and region based convolutional neural network of sample size 22. The number plate of the vehicle will be detected and converted into string format. Results: A prediction accuracy of 96.4% using the optical character recognition method was achieved, while the region based convolutional neural network was 94.2.%.The significance value obtained in statistical analysis is 0.02(p<0.05). Conclusion: The results show that identification of number plates is significantly better in optical character recognition than region based convolutional neural networks.