Detecting Specific Building Structure in Videos Using Deep Learning.

Abstract

Data is growing at a colossal rate with the advancement of technology. Video data has massively increased as well, which comprises of images, videos, and audios. The digital videos are sequence of images containing a lot of information. The process of searching is tedious and time consuming, not to mention the amount of time it generally takes to search for something specific. Searching in images and videos is especially challenging. Therefore, there is a need for a system that can effectively detect and recognize certain building object(s) in videos. The idea behind our application is to make content based searching easier and to provide a system that can accurately detect specific building structures from a wide range of videos. We have chosen six famous buildings of Pakistan for detection. These six famous structure are, Faisal Mosque, Minar e Pakistan, Pakistan Monument, Centaurus, Parliament Building and Bahria University Islamabad Campus. The system detects these specific structures with a precision of 93.75 percent and a recall of 89.05 percent. The system using Convolutional Neural Network CNN to accurately detect different structures, a bounding box is drawn around the detected structures and confidence score is displayed