Sergio Izquierdo
@sizquierdo.bsky.social
PhD candidate at University of Zaragoza.
Previously intern at Niantic Labs and Skydio.
Working on 3D reconstruction and Deep Learning.
serizba.github.io
Previously intern at Niantic Labs and Skydio.
Working on 3D reconstruction and Deep Learning.
serizba.github.io
We focused on depth from videos and as you pointed we didn't train on datasets with different captures per scene.
March 31, 2025 at 3:51 PM
We focused on depth from videos and as you pointed we didn't train on datasets with different captures per scene.
Check the website: nianticlabs.github.io/mvsanywhere/
And the paper: arxiv.org/pdf/2503.22430
Code coming soon!
Great work with @mohamedsayed.bsky.social @mdfirman.bsky.social @guiggh.bsky.social D. Turmukhambetov @jcivera.bsky.social @oisinmacaodha.bsky.social @gbrostow.bsky.social J. Watson
And the paper: arxiv.org/pdf/2503.22430
Code coming soon!
Great work with @mohamedsayed.bsky.social @mdfirman.bsky.social @guiggh.bsky.social D. Turmukhambetov @jcivera.bsky.social @oisinmacaodha.bsky.social @gbrostow.bsky.social J. Watson
MVSAnywhere: Zero-Shot Multi-View Stereo
MVSAnywhere: Zero-Shot Multi-View Stereo, CVPR 2025
nianticlabs.github.io
March 31, 2025 at 12:52 PM
Check the website: nianticlabs.github.io/mvsanywhere/
And the paper: arxiv.org/pdf/2503.22430
Code coming soon!
Great work with @mohamedsayed.bsky.social @mdfirman.bsky.social @guiggh.bsky.social D. Turmukhambetov @jcivera.bsky.social @oisinmacaodha.bsky.social @gbrostow.bsky.social J. Watson
And the paper: arxiv.org/pdf/2503.22430
Code coming soon!
Great work with @mohamedsayed.bsky.social @mdfirman.bsky.social @guiggh.bsky.social D. Turmukhambetov @jcivera.bsky.social @oisinmacaodha.bsky.social @gbrostow.bsky.social J. Watson
💡Use case:
We show how the accurate and robust depths from MVSAnywhere serve to regularize gaussian splats, obtaining much cleaner scene reconstructions.
As MVSAnywhere is agnostic to the scene scale, this is plug-and-play for your splats!
We show how the accurate and robust depths from MVSAnywhere serve to regularize gaussian splats, obtaining much cleaner scene reconstructions.
As MVSAnywhere is agnostic to the scene scale, this is plug-and-play for your splats!
March 31, 2025 at 12:52 PM
💡Use case:
We show how the accurate and robust depths from MVSAnywhere serve to regularize gaussian splats, obtaining much cleaner scene reconstructions.
As MVSAnywhere is agnostic to the scene scale, this is plug-and-play for your splats!
We show how the accurate and robust depths from MVSAnywhere serve to regularize gaussian splats, obtaining much cleaner scene reconstructions.
As MVSAnywhere is agnostic to the scene scale, this is plug-and-play for your splats!
🏆Results:
MVSAnywhere achieves state-of-the-art results on the Robust Multi-View Depth Benchmark, showing its strong generalization performance.
MVSAnywhere achieves state-of-the-art results on the Robust Multi-View Depth Benchmark, showing its strong generalization performance.
March 31, 2025 at 12:52 PM
🏆Results:
MVSAnywhere achieves state-of-the-art results on the Robust Multi-View Depth Benchmark, showing its strong generalization performance.
MVSAnywhere achieves state-of-the-art results on the Robust Multi-View Depth Benchmark, showing its strong generalization performance.
🧩Challenge: Varying Depth Scales & Unknown Ranges
🔹Most models require a known depth range to estimate the cost volume.
✅MVSAnywhere estimates an initial range based on camera scale and setup and refines it. It predicts at the same scale as the input cameras!
🔹Most models require a known depth range to estimate the cost volume.
✅MVSAnywhere estimates an initial range based on camera scale and setup and refines it. It predicts at the same scale as the input cameras!
March 31, 2025 at 12:52 PM
🧩Challenge: Varying Depth Scales & Unknown Ranges
🔹Most models require a known depth range to estimate the cost volume.
✅MVSAnywhere estimates an initial range based on camera scale and setup and refines it. It predicts at the same scale as the input cameras!
🔹Most models require a known depth range to estimate the cost volume.
✅MVSAnywhere estimates an initial range based on camera scale and setup and refines it. It predicts at the same scale as the input cameras!
🧩Challenge: Domain Generalization
🔹Previous models struggle across different domains ( indoor🏠 vs outdoor🏞️).
✅MVSAnywhere uses a transformer architecture and is trained on a large array of varied synthetic datasets
🔹Previous models struggle across different domains ( indoor🏠 vs outdoor🏞️).
✅MVSAnywhere uses a transformer architecture and is trained on a large array of varied synthetic datasets
March 31, 2025 at 12:52 PM
🧩Challenge: Domain Generalization
🔹Previous models struggle across different domains ( indoor🏠 vs outdoor🏞️).
✅MVSAnywhere uses a transformer architecture and is trained on a large array of varied synthetic datasets
🔹Previous models struggle across different domains ( indoor🏠 vs outdoor🏞️).
✅MVSAnywhere uses a transformer architecture and is trained on a large array of varied synthetic datasets
🧩Challenge: Robustness to casually captured videos
🔹MVS methods completely rely on the matches of the cost volume (not working for low overlap & dynamic)
✅MVSAnywhere successfully combines strong single-view image priors with multi-view information from our cost volume
🔹MVS methods completely rely on the matches of the cost volume (not working for low overlap & dynamic)
✅MVSAnywhere successfully combines strong single-view image priors with multi-view information from our cost volume
March 31, 2025 at 12:52 PM
🧩Challenge: Robustness to casually captured videos
🔹MVS methods completely rely on the matches of the cost volume (not working for low overlap & dynamic)
✅MVSAnywhere successfully combines strong single-view image priors with multi-view information from our cost volume
🔹MVS methods completely rely on the matches of the cost volume (not working for low overlap & dynamic)
✅MVSAnywhere successfully combines strong single-view image priors with multi-view information from our cost volume