IDESplat

<p align="center"> <h1 align="center">IDESplat: Iterative Depth Probability Estimation for Generalizable 3D Gaussian Splatting</h1> <p align="center"> <a href="https://scholar.google.com/citations?user=CsVTBJoAAAAJ&hl=zh-CN">Wei Long</a> · <a href="https://scholar.google.com/citations?user=rvVphXoAAAAJ&hl=zh-CN&oi=ao">Haifeng Wu</a> · <a href="https://openreview.net/profile?id=~Shiyin_Jiang1">Shiyin Jiang</a> · <a href="https://scholar.google.com/citations?user=tyYxiXoAAAAJ&hl=zh-CN">Jinhua Zhang</a> · <a href="https://openreview.net/profile?id=~Xinchun_Ji2">Xinchun Ji</a> · <a href="https://scholar.google.com/citations?user=-kSTt40AAAAJ&hl=zh-CN">Shuhang Gu</a> </p> <h3 align="center">CVPR 2026</h3> <h3 align="center"><a href="https://arxiv.org/abs/2601.03824">Paper</a> | <a href="https://github.com/CVL-UESTC/IDESplat">Code</a></h3> <div align="center"></div> </p> <p align="center"> <a href=""> <img src="figures/IDESplat_m.png" alt="IDESplat Overview" width="100%"> </a> </p>

Architecture

<p align="center"> <a href=""> <img src="figures/IDESplat_net.png" alt="IDESplat Overview" width="100%"> </a> </p>

The overall architecture of IDESplat. IDESplat is composed of three key parts: a feature extraction backbone, an iterative depth probability estimation process, and a Gaussian Focused Module (GFM). The iterative process consists of cascaded Depth Probability Boosting Units (DPBUs). Each unit combines multi-level warp results in a multiplicative manner to mitigate the inherent instability of a single warp. As IDESplat iteratively updates the depth candidates and boosts the probability estimates, the depth map becomes more precise, leading to accurate Gaussian means.

Comparison of depth maps:

<p align="center"> <a href=""> <img src="figures/visual_depth_map.png" alt="IDESplat Overview" width="100%"> </a> </p>