Robust rotation estimation for arbitrary video sequences is useful for a host of downstream tasks. Check out our work with Aalok Patwardhan, Gwangbin Bae.

How can we infer 3D-consistent poses and dense geometry in real-time given only RGB images? 𝗖𝗢𝗠𝗢 decodes dense geometry from a compact and optimizable set of 3D anchor points to enforce 3D consistency. Project page: edexheim.github.io/como/ Work with Andrew Davison 1/n

Our code is now available! Webcam feeds, video files, random YouTube videos -- you name it and you can try to get camera orientation from it 🤳 It works surprisingly well even in non-Manhattan environments so go download it and break it

📢 Code release for 𝗗𝗦𝗜𝗡𝗘 (#CVPR2024 - Oral) DSINE gives you surface normal prediction (+ uncertainty) in real-time. We have released the code for training, testing, and running real-time demos. Try it yourself! github.com/baegwangbin/DS…

We'll be presenting "Distributed Simultaneous Localisation and Auto-Calibration using GBP" at #IROS2024 We show that distributed GBP can be effectively used to localise and auto-calibrate many robots. Please come to our presentation (Wed 17:30 Room 11) if you're around!

Introducing MASt3R-SLAM, the first real-time monocular dense SLAM with MASt3R as a foundation. Easy to use like DUSt3R/MASt3R, from an uncalibrated RGB video it recovers accurate, globally consistent poses & a dense map. With Eric Dexheimer*, Andrew Davison (*Equal Contribution)