LDCM: Large Depth Completion Model from Sparse Observations

📝 Introduction

LDCM is a high-performance depth completion framework. It effectively reconstructs high-fidelity dense metric depth maps from a single RGB image and sparse depth measurements.

---

🛠 Roadmap

• Release project paper and technical details.
• Release pre-trained models.
• Release training and evaluation scripts.
• Support for spatio-temporal video depth completion.