Real-Time Compression Artifact Reduction for Low Bandwidth Video Conferencing
Ajinkya Kudoo, Aman Pahurkar, Umang Prajapati, Jyoti Ramteke · 2021
Video conferencing has been proved to be beneficial in terms of productivity, saving time and travel expenses as well. People have relied on video conferencing for business and academic purposes and the current pandemic situation has increased its usage. However, many people face issues related to low internet bandwidth. In order to transmit images or videos over the available low bandwidth, they are compressed so that take less space and can be sent over even on low bandwidth. Low bandwidth adds compression artifacts to transmitted media. Many systems have proposed various deep learning approaches to reduce these artifacts. However, they are not suited for real-time usage. We are proposing a solution to reduce compression artifacts by utilizing a deep learning model which will reduce compression artifacts in real time to improve video conferencing experience.