AI-based Saliency Prediction for better perceptual quality in Display and Video Compression

This research project aims to develop AI based models for predicting visual saliency in gaming videos. Visual saliency refers to identifying the most relevant and important areas in an image or video. This technology has numerous applications, including improving video compression and optimizing the backlight dimming of displays, leading to better user experience and power savings. However, there is currently a lack of saliency prediction datasets for gaming videos, which are essential for developing accurate models. This project will explore ways to enrich these datasets and improve the accuracy of the models using advanced machine learning techniques. The partner organization, AMD, will benefit from the research by gaining expertise in video saliency technology, particularly for gaming, which can be used to create more energy-efficient displays and better video compression, leading to improved user experience.

Faculty Supervisor:

Qiang Sun

Student:

Partner:

AMD Canada

Discipline:

Computer science

Sector:

Manufacturing; Professional, scientific and technical services

University:

University of Toronto

Program: