This work focuses on using these feature-based sparse patterns to generate additional depth information by interpolating regions between clusters of samples that are in close proximity to each other.
Training a single network for high resolution and geometrically consistent monocular depth estimation is challenging due to varying scene complexities in the real world.
During this project, state-of-the-art deep learning models have been used to estimate depth maps from a monocular RGB image applying a teacher-student learning approach.
This is an ideal reference for anyone looking for an understanding of the diverse concepts and ideas around this topic and how we can move towards more general techniques than traditional photometric stereo.
This thesis evaluates and profiles a monocular depth estimation algorithm in which depth maps are generated from a single image using a non-parametric depth transfer approach. 3D depth from images has a wide range of applications in ...
... metrics Kendall's τ and WHDR. Table 4 shows the effectiveness of the two-phase training scheme of the proposed algorithm. The proposed algorithm ... Depth Map Decomposition for Monocular Depth Estimation 31 4.6 Analysis 5 Conclusions.