This proposal provides a technique to change an online video meeting background in real-time using deep learning. By changing the background video for participants of a meeting, the participants can protect their accuracy. The background-changing technique of this proposal may include a deep learning model for video semantic segmentation. One or more Spatio-Temporal Transformer Gated Recurrent Units (STGRUs) may be utilized to enhance classification accuracy and to add an optical flow into the model; thereby increasing calculation speed.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.