Abstract:
Videos are dynamic in nature. They contain pictures as well as text with different dimensions and colors per frame. Videos contain valuable information in the form of text. The salient features of text in videos are the primary video items. The purpose is to extract these salient entities which are useful in analysis of text. Text analysis in videos is all in one product which analyze text in a video using sentiment analysis techniques. Different modules like detection of text, recognition of text, its analysis and in the end indexing feature multimedia content requires a combination of various artificial intelligence techniques. This project will analyze text sentiments in video beforehand and censor the video for containing negative sentiments. It differentiate textual content from non-textual content in videos. It is capable of extracting text from colored backgrounds or images that have complex patterns present in them. Despite of text coming in oblique viewing angles with low resolution and having some kind of noise or distortions, it can recognize text in good quality. It provides high accuracy by combining text information that appears over multiple video frames.
