MxSR: Monocular Depth Estimation Cross-Feat Stereo Depth Estimation with Refinement

MxSR is an advanced depth estimation framework that combines the strengths of monocular depth estimation and stereo depth estimation through a cross-feature learning approach, leveraging a powerful refinement network. The goal is to achieve high-quality, accurate depth maps from both monocular and stereo inputs, improving the robustness and precision of depth perception.