Automated Video Quality Analysis Using Deep Learning
Yashas Dewan, G. Vadivu, Mihir Singh Malik · 2024
Video streaming is a common service utilized in today's times, ranging from popular platforms such as Netflix, Disney, and Amazon Prime to live video calls facilitated by WhatsApp and FaceTime. Given the growing computational capabilities and the imperative of delivering optimal end-user experiences, there arises a necessity to develop methods for testing and analyzing the video quality of streaming content. In our research, we propose a method for analyzing the same as well as a substantial publicly available dataset comprising thousands of distorted frames. We employ a Convolutional Neural Network (CNN) to train a model for discerning whether a frame in a video input is glitched or not, addressing the imperative of ensuring seamless video playback and enhanced user satisfaction.