Fangjinhua Wang
@fangjinhuawang.bsky.social
Postdoc at ETH zurich. Previous PhD at ETH Zurich. Previous intern at Google, Meta, Microsoft. Homepage: fangjinhuawang.github.io
[T-PAMI 2025] Introducing a lightweight and accurate framework for 3D reconstruction utilizing diffusion model! While there is often a tradeoff between accuracy and efficiency, we believe we have achieved an optimal balance.
Code: github.com/cvg/diffmvs. Try with your own images!
Code: github.com/cvg/diffmvs. Try with your own images!
September 3, 2025 at 7:49 PM
[T-PAMI 2025] Introducing a lightweight and accurate framework for 3D reconstruction utilizing diffusion model! While there is often a tradeoff between accuracy and efficiency, we believe we have achieved an optimal balance.
Code: github.com/cvg/diffmvs. Try with your own images!
Code: github.com/cvg/diffmvs. Try with your own images!
#CVPR2025 Thrilled to present our latest work on visual localization at Nashville next week! In R-SCoRe, we close the gap between Scene Coordinate Regression (SCR) and feature matching methods on challenging benchmarks with a small map.
Paper: arxiv.org/pdf/2501.01421
Code: github.com/cvg/scrstudio
Paper: arxiv.org/pdf/2501.01421
Code: github.com/cvg/scrstudio
June 5, 2025 at 1:44 PM
#CVPR2025 Thrilled to present our latest work on visual localization at Nashville next week! In R-SCoRe, we close the gap between Scene Coordinate Regression (SCR) and feature matching methods on challenging benchmarks with a small map.
Paper: arxiv.org/pdf/2501.01421
Code: github.com/cvg/scrstudio
Paper: arxiv.org/pdf/2501.01421
Code: github.com/cvg/scrstudio