English

PureCLIP-Depth: Prompt-Free and Decoder-Free Monocular Depth Estimation within CLIP Embedding Space

Computer Vision and Pattern Recognition 2026-03-18 v1

Abstract

We propose PureCLIP-Depth, a completely prompt-free, decoder-free Monocular Depth Estimation (MDE) model that operates entirely within the Contrastive Language-Image Pre-training (CLIP) embedding space. Unlike recent models that rely heavily on geometric features, we explore a novel approach to MDE driven by conceptual information, performing computations directly within the conceptual CLIP space. The core of our method lies in learning a direct mapping from the RGB domain to the depth domain strictly inside this embedding space. Our approach achieves state-of-the-art performance among CLIP embedding-based models on both indoor and outdoor datasets. The code used in this research is available at: https://github.com/ryutaroLF/PureCLIP-Depth

Keywords

Cite

@article{arxiv.2603.16238,
  title  = {PureCLIP-Depth: Prompt-Free and Decoder-Free Monocular Depth Estimation within CLIP Embedding Space},
  author = {Ryutaro Miya and Kazuyoshi Fushinobu and Tatsuya Kawaguchi},
  journal= {arXiv preprint arXiv:2603.16238},
  year   = {2026}
}

Comments

12 pages, 4 figures

R2 v1 2026-07-01T11:23:46.029Z