HistReNeRF framework relocalizes historic photos in 3D scenes

Researchers have developed HistReNeRF, a novel framework designed to accurately relocalize historical photographs within contemporary 3D scene reconstructions. This method addresses the challenges posed by differences in appearance, objects, and spatial layouts between historical and modern imagery. By adapting DINOv2 patch features and querying a Neural Radiance Field (NeRF) reconstruction, HistReNeRF estimates the 6-DoF pose of historical photos. Evaluations on a new dataset of European landmarks demonstrated that this embedding-space adaptation reduces translation and rotation errors by an average of 11% and 16%, respectively, outperforming pixel-space methods. AI

IMPACT Enhances historical photo analysis and 3D scene reconstruction capabilities.

RANK_REASON The item is a research paper published on arXiv detailing a new method for image relocalization. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• DINOv2
• Europe
• Gotit.pub
• HistReNeRF
• Hugging Face
• Nerf
• Neural radiance field
• ScienceCast

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →