Researchers have developed PaGeR, a framework that adapts existing 3D foundation models, originally designed for perspective images, to reconstruct full 360-degree scenes from single panoramic images. This approach allows for a unified, single-pass estimation of scale-invariant depth, metric depth, surface normals, and sky masks. By minimizing architectural changes and training with a mix of perspective and panoramic data, PaGeR retains the underlying model's 3D prior while enabling consistent 360-degree scene estimation, achieving state-of-the-art performance. AI
IMPACT Enables reconstruction of full 360-degree scenes from single images, potentially advancing applications in robotics, VR, and autonomous systems.
RANK_REASON The cluster describes a new research paper detailing a novel framework for 3D geometry estimation.
Read on Hugging Face Daily Papers →
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
