A neural network based video playback system

Video playback system development has always been a challenging task in digital world problems such as education, sports event coverage, news and media domains. This work introduces a novel approach to summarize videos, primarily video lectures by varying the video playback frame rate without removing any frames from the video. This is accomplished by using Self organizing feature maps so as to learn from previous users' data and give the optimized playback frame rate as the output. A video player is modified to record the key strokes and create a database of previous users' playback preferences. Being a self-learning system, the accuracy of results is better when training is done using a vast set of data. Finally, a comparison of the original video and the final output video is made by analyzing the experience of volunteers.