Introducing “FlowMap”, the first self-supervised, differentiable structure-from-motion method that is competitive with conventional SfM like Colmap!

cameronosmith.github.io/flowmap/

IMO this solves a major missing piece for internet-scale training of 3D Deep Learning methods.

1/n

On the path to AGI, we scrambled letters, and somehow ended up on GAI (G=generative)😅

The keynotes from #3DV2024 are now publicly available! youtube.com/@3DV2024

International Conference on 3D Vision for broadcasting

I added a Google Colab for anyone interested in quickly finding a set of common keypoints on a set of custom images colab.research.google.com/drive/1ULz-HIR…

🚨🇨🇦🏔️You heard it here first - Wayve is coming to Canada and so am I!
I’m thrilled to be opening our office in Vancouver and to be working alongside the world-renowned computer vision and ML researcher Yasutaka Furukawa who's just joined us.
The team is ramping up fast, and we…

Hold your horses, something big is coming this weekend (or Monday... according to arXiv's mood)

Just some🥰 and 🥳 to PhD collaborators Lily Goli and Eric Hedlin for theirs 𝐬𝐩𝐨𝐭𝐥𝐢𝐠𝐡𝐭𝐬, and to David Charatan for his 𝐨𝐫𝐚𝐥 at #CVPR2024

Project pages:
davidcharatan.com/pixelsplat (oral)
bayesrays.github.io (spotlight)
ubc-vision.github.io/StableKeypoints (spotlight)

Code for T4CD is public!

Code is out for TC4D—a method for controllable 4D gen that animates objects along trajectories—check it out!
sherwinbahmani.github.io/tc4d

w/ Sherwin Bahmani Xian Liu Yifan Wang Ivan Skorokhodov Victor Rong Ziwei Liu Xihui Liu Jeong Joon Park Sergey Tulyakov Gordon Wetzstein Andrea Tagliasacchi 🇨🇦🏔️

David & Lester have just updated the pixelSplat code, now supporting 3-view training in addition to 2-view and with some improvements suggested to us by reviewers, including new pre-trained checkpoints!
davidcharatan.com/pixelsplat/

More exciting work coming very soon, stay tuned!

🫣