Computer vision for immersive imagery
ImmersiveDepth
Monocular depth estimation for 360° imagery.
A hybrid depth-estimation workflow combining tangent projection and multiple models, published at IEEE VRW 2025.
Approach
Research approach.
The workflow takes a panoramic image through projection, depth estimation, post-processing and alignment, leading to a three-dimensional visualisation. The IEEE VRW paper describes the method and its integration; the accepted manuscript is available through Bournemouth University.
This computer-vision project connects to my interest in constructing immersive environments. Working with panoramic imagery lets me examine how a captured scene can be represented spatially, bringing image processing closer to the experience of viewing a three-dimensional environment.
Related research