A weighted fusion of GLPN and Depth-Anything depth maps plus Poisson meshing reconstructs 3D artwork models from single images and beats either depth model alone on CLIP similarity for five test paintings.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Augmented Reality in Cultural Heritage: A Dual-Model Pipeline for 3D Artwork Reconstruction
A weighted fusion of GLPN and Depth-Anything depth maps plus Poisson meshing reconstructs 3D artwork models from single images and beats either depth model alone on CLIP similarity for five test paintings.