Michael Firman retweeted
Splats are great, but wouldn’t it be wonderful if we could transform them into the stuff of dreams? Introducing our #CVPR2025 paper ✨Morpheus✨!
Morpheus lets you change the shape and appearance of 3D Gaussian splats with a few words!
nianticlabs.github.io/morphe…
(1/6)
Michael Firman retweeted
We have released the code and model weights on github.com/nianticlabs/HandD… #ECCV2024
👋Would you like more accurate 3D camera-space hand mesh predictions? Check our new #ECCV2024 work: "HandDGP: Camera-Space Hand Mesh Prediction with Differentiable Global Positioning". Project page: nianticlabs.github.io/handdg… (1/5)
Michael Firman retweeted
This election in South East Cornwall is about change.
It’s our chance for our special corner of Cornwall to get its fair share from the next Government. From a Labour government.
We need a fresh start.
Michael Firman retweeted
DoubleTake: Geometry Guided Depth Estimation
@MohammedAmr1 @AleottiFilippo, Jamie Watson, Zawar Qureshi, @gui_ggh, Gabriel Brostow, Sara Vicente, @_mdfirman
tl;dr: SimpleRecon + "hint MLP", processing cost volume + previous depth estimate.
arxiv.org/abs/2406.18387
Michael Firman retweeted
"AirPlanes: Accurate Plane Estimation via 3D-Consistent Embeddings"
Jamie Watson, @AleottiFilippo, @MohammedAmr1, Zawar Qureshi, @oisinmacaodha, Gabriel Brostow, @_mdfirman, Sara Vicente
Poster WED-PM: PS-2-42
nianticlabs.github.io/airpla…
Michael Firman retweeted
Today's PUNCH colour cover. What larks! Summer frolics, medieval style. Title page, Summer Number 1929 by George Morrow #history #MiddleAges #leisure #fun #holidays #vacations #seaside #piers #sport #cricket #tennis #golf #manuscripts
Michael Firman retweeted
Come see us at our poster today in Vancouver; I promise you shiny videos!
WED-AM-079
Were you ever frustrated with jagged edges in AR?! 💔 No more! ✨
📢 Introducing our #CVPR2023 paper (and code!):
Virtual Occlusions Through Implicit Depth❤️🎉
nianticlabs.github.io/implic…
arxiv.org/abs/2305.07014
[1/10]
Michael Firman retweeted
We continue on Wednesday morning, with showing how to learn fine-grained occlusion masks. Poster WED-AM-079
Work by Jamie Watson, @MohammedAmr1 , Zawar Qureshi, Gabriel J. Brostow, Sara Vicente, Oisin Mac Aodha, @_mdfirman
teapawt.org/MohammedAmr1/status/16…
Were you ever frustrated with jagged edges in AR?! 💔 No more! ✨
📢 Introducing our #CVPR2023 paper (and code!):
Virtual Occlusions Through Implicit Depth❤️🎉
nianticlabs.github.io/implic…
arxiv.org/abs/2305.07014
[1/10]
Michael Firman retweeted
We end with a bang on Thursday, poster THU-AM-005, by showing how to remove objects from NeRFs.
Work by @weder_silvan , @gui_ggh , @AMonszpart , @mapo1 , Gabriel Brostow, @_mdfirman , Sara Vicente
Michael Firman retweeted
Niantic research has a strong presence at #CVPR2023 with 5 papers (one highlight) and various contributions throughout workshops and tutorials.
Diffusion, NeRF, relocalisation, object pose, feature matching, depth and occlusions.
Here is where you can catch us:
Michael Firman retweeted
Does the #ACErelocalizer work? I mean... really!? In practise!?
You bet it does. We will officially present it this June in Vancouver, but it has been deployed as part of the @LightshipAR visual position system from day one. At work for more than 1 year.
teapawt.org/eric_brachmann/status/…
#LightshipVPS in action. Please note the alignment of the chain with the anchor top 🙀
@LightshipAR @NianticEng #VPS #AR #AugmentedReality
Michael Firman retweeted
Camera Pose Estimation and Localization
with Active Audio Sensing: ecva.net/papers/eccv_2022/pa…
Can audio help camera pose estimation? New ECCV paper shows SuperGlue+Audio helps in wide baseline and low light scenarios. Awesome work from @mrharicot @eric_brachmann @_mdfirman et al
Michael Firman retweeted
Michael Firman retweeted
📢SimpleRecon will be presented at #ECCV2022!
Although I'd love to come, I can't be at the conference due to a restrictive visa process. My coauthors will be there, so you'll be in good hands!
Meanwhile you'll have me explaining our paper in video form: youtu.be/3LP8jp45Ef8
Michael Firman retweeted
Today, we launch Map-Free Visual Relocalisation. A twist on the usual relocalisation formula. A new dataset. A new benchmark.
Check out our video! storage.cloud.google.com/nia…
#MapFreeReloc to be presented at #ECCV2022
Michael Firman retweeted
Today I’m finally sharing the news I’ve bursting with for over a year!
The Grove: A Nature Odyssey in 19 ½ Front Gardens will be published by @Octopus_Books on the 7th of April. I’m so proud. It really is the best thing I’ve ever done and I can’t wait for you to read it
Michael Firman retweeted
Wondering how to detect when your neural network is about to predict pure non-sense in a safety critical scenario?
We answer your questions in our #ICCV2021 @ICCV_2021 paper!
Thursday 1am (CET) or Friday 6pm (CET), Session 12, ID: 3734
📜 openaccess.thecvf.com/conten…
Thread 🧵👇
Michael Firman retweeted
IMHO, these claims are bizarre.
Here are four 2D embeddings of MNIST: UMAP (aka t-SNE with exaggeration); PCA; untrained network; the elephant 🐘.
Chari et al. claim that the latter three are as good as or better than UMAP -- even for local structure preservation. Really? [5/n]
Michael Firman retweeted
Scanning with laser/lidar has a better geometry but the textures are not enough for #VFX and #gamedev . In @RealityCapture_ you can mix #lidar+ #photogrammetry for better textures. Next step, import into #UE5 :-)
























