Semantic Analysis of News Based on the Deep Convolution Neural Network

Abstract

Video data analysis is a fascinating field since the last few decades. It assists user to find the genre of a video without watching it. In this paper, an application for analysing Pakistani news data based on the scene classification using deep convolution neural network has been presented. For this purpose the dataset has been collected on our own, which consists of 200 videos of different news channels, and covers almost all categories which we possess in our methodology. The results have been evaluated using the 2D convolution neural network on the fine-tuned inception model. The methodology achieved the 92.2% accuracy on the proposed architecture, such a high accuracy on locally prepared dataset of the news for the video data analysis shows the novelty in literature.