SENTENCE FRAMING USING VIDEO

Abstract

To cope with the large amount of online videos we downloaded from YouTube, we transform every video shot into
a set of images by extracting the video shot to interpret the videos for better understanding the content that it holds. Here we
convert the video into query images and store it as testing dataset. In this phase, we make use of Gabor feature and wavelet
transform on the query image cluster extracted from the query video shot V to extract the various features out of it.
Visualization of the Gabor Filter for edge detection and extraction of texture for simple and complex cells. Wavelet transfer
uses low pass filter and high pass filter to detect the various edges and textures. Extracting the features of the training image
and these the features are stored in the array format. We extract features for the testing data set also. We further make use of
machine learning to compare and find out perfect match of query image. The data set plays important role. In these step we
making use of machine learning’s neural network wherein we give testing data set as in input node and compared with the
nodes to give the correct output. The best matched image is then compared with the various classes in order to extract the
keyword respectevely. The keywords are then classified mainly as per the three domains namely object, event, scenes. We
then frame a sentence using acquired keyword.