HQ-OV3D combines VLM-derived proposals with a diffusion denoiser that transfers box geometry from base classes to improve open-vocabulary 3D detection.
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
REJECT 1representative citing papers
citing papers explorer
-
HQ-OV3D: A High Box Quality Open-World 3D Detection Framework based on Diffision Model
HQ-OV3D combines VLM-derived proposals with a diffusion denoiser that transfers box geometry from base classes to improve open-vocabulary 3D detection.