Analyzing Video to Find Particular Object Timestamp

  • Sufiyan Sayyed et al.

Abstract

We all are surrounded by huge data. People upload, download, send and update video,
audio, images, and documents from a variety of devices. We often need to find one particular item of
data among these hundreds, thousands, millions or more of these data objects. In documents or onweb page we’ve all gotten the little magnifying glass to bring up the search field. So we type the
keyword in the search field and immediately get a list of every time that word shows up. Nowadays
we can also search in images by using object detection. But searching in videos is currently not
feasible. As more and more information gets in a large amount of which is left unprocessed.
Video by itself is really hard to search. We can’t find videos we want or browse for what we
might like. That’s a loss. So in our project, we provide a basic solution for it, so that we are able to
find a particular object in videos and return the timestamp, where that object-related image was
encountered. In the proposed system user will provide video as well as tag or image for the object to
search. Then we process the video frame by frame and keep track of continuity of tag in frames. After
the parsing of the entire video, we will provide a timestamp when particular tag related object
encountered in the entire video.

Published
2020-05-20