PR-IQA creates geometrically consistent partial quality maps from reference overlaps and completes them via cross-attention to evaluate diffusion-synthesized views without ground truth, achieving full-reference accuracy and improving 3DGS reconstructions.
Training Data Generation Frame Sampling.We utilize the Map-free Visual Relo- calization (MFR) dataset [1] as our primary source
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
PR-IQA: Partial-Reference Image Quality Assessment for Diffusion-Based Novel View Synthesis
PR-IQA creates geometrically consistent partial quality maps from reference overlaps and completes them via cross-attention to evaluate diffusion-synthesized views without ground truth, achieving full-reference accuracy and improving 3DGS reconstructions.