Towards Automating Video Inpainting for Use in a 2D to 3D Video Conversion Pipeline

The goal of the project is to reduce the workload involved in the occlusion filling step of a 2D to 3D video conversion pipeline. To do so, the intern will develop and implement a video inpainting algorithm suitable for use in Hollywood Films. The algorithm will be integrated into the organization’s 2D to 3D video conversion pipeline. The algorithm will provide controls which allow the user to manipulate the results of the inpainting process. The intern will experiment with the use of additional data available within the pipeline as a means to overcome issues with current video inpainting methods. In particular, changing illumination and handling of fixed light sources will be addressed. Since inpainting can be extremely tedious and time consuming when performed by hand, a more automated solution will greatly increase the organization’s efficiency in performing 2D to 3D conversions.

Faculty Supervisor:

Amy Gooch

Student:

Partner:

Discipline:

Computer science

Sector:

University:

University of Victoria

Program:

Accelerate