Analysis of region of interest (RoI) of multimedia content using eye-tracker: poster

With the advent of smart-phones and YouTube, viewing video and multimedia content over wireless medium has gained enormous interest. In this regard, encoding the video with high-quality encoder results in humongous increase in the data-rate. Given that users tend to focus more on certain parts of video, the region of video frame that is highly viewed is encoded with best-possible quality and the other regions are encoded with decreasing quality. However, the main challenge is to ensure that adaptive encoding does not result in significant perceivable degradation in video quality. In this 2-pager for poster, the nature of region of interest (RoI) is analyzed by identifying the user's focus while watching video captured with two different camera technology. It has been observed that user's RoI follows a Gaussian distribution from left to right of the screen. At the same time, the user's RoI from top to bottom follows a distribution that is almost Gaussian, but varies with the nature of the video content. This is a significant result that will play a key role in designing an adaptive RoI for reducing the data rate while transmitting multimedia content over the network.