Sign in

Johan Edstedt

@parskatt.bsky.social
1.7K followers 422 following 2.1K posts

PhD student @ Linköping University I like 3D vision and training neural networks. Code: github.com/parskatt Weights: github.com/Parskatt/storage/release…

PostsRepliesMedia
Johan Edstedt @parskatt.bsky.social · 24/07/2026
Recently read an environmental warning (PFAS mainly) about eating fish caught from some nearby lakes. I was curious so I checked the source:
110
Johan Edstedt @parskatt.bsky.social · 20/05/2026
190
Johan Edstedt @parskatt.bsky.social · 19/05/2026
Functional programming be like:
050
Johan Edstedt @parskatt.bsky.social · 11/05/2026
Me to my models in wandb
080
Johan Edstedt @parskatt.bsky.social · 17/04/2026
Steer!
050
Johan Edstedt @parskatt.bsky.social · 15/04/2026
1140
Johan Edstedt @parskatt.bsky.social · 15/04/2026
How are these so accurate
120
Johan Edstedt @parskatt.bsky.social · 15/04/2026
Oof
030
Johan Edstedt @parskatt.bsky.social · 14/04/2026
Accepted to the Image Matching Workshop at #CVPR26!
1171
Johan Edstedt @parskatt.bsky.social · 14/04/2026
Introducing LoMa-R! A rotation invariant version of LoMa. Code: github.com/davnords/loma Paper: arxiv.org/abs/2604.11809
1183
Johan Edstedt @parskatt.bsky.social · 13/04/2026
Checking my training batch
0151
Johan Edstedt @parskatt.bsky.social · 07/04/2026
0211
Johan Edstedt @parskatt.bsky.social · 07/04/2026
Introducing HardMatch!
1120
Johan Edstedt @parskatt.bsky.social · 07/04/2026
Introducing LoMa, the next generation of feature matcher!
3414
Johan Edstedt @parskatt.bsky.social · 06/04/2026
from loma import LoMa
050
Johan Edstedt @parskatt.bsky.social · 06/04/2026
090
Johan Edstedt @parskatt.bsky.social · 06/04/2026
Out in a few hours!
090
Reposted by Johan Edstedt
Christian Wolf @chriswolfvision.bsky.social · 30/03/2026
The keynotes and award talks at 3DV 2026 have been made available, with talks by Christian Rupprecht ("3D computer vision done and Dusted?", Angela Dai, Alec Jacobson, Jitendra Malik, Songyou Peng and Michael Niemeyer. www.youtube.com/playlist?lis...
0226
Johan Edstedt @parskatt.bsky.social · 28/03/2026
Had a shampoo that bragged about having 92% natural ingredients. Checked the back and it was 89% water.
2110
Johan Edstedt @parskatt.bsky.social · 20/03/2026
Make sure to come by our poster right now if you are at #3DV2026!
181
Johan Edstedt @parskatt.bsky.social · 15/03/2026
2132
Reposted by Johan Edstedt
David Nordström @davnords.bsky.social · 13/03/2026
Yesterday, @parskatt.bsky.social and the RoMa v2 team got an award at the annual Swedish Symposium on Image Analysis. FeelsGoodMan. @bokmangeorg.bsky.social @fredkahl.bsky.social @jastermark.bsky.social @liucvl.bsky.social et. al.
3143
Johan Edstedt @parskatt.bsky.social · 07/03/2026
Coming soon
1190
Johan Edstedt @parskatt.bsky.social · 06/03/2026
Putting a vowel before 3R should be illegal.
330
Johan Edstedt @parskatt.bsky.social · 05/03/2026
090
Johan Edstedt @parskatt.bsky.social · 05/03/2026
0230
Johan Edstedt @parskatt.bsky.social · 05/03/2026
1130
Johan Edstedt @parskatt.bsky.social · 06/02/2026
180
Reposted by Johan Edstedt
Johan Nohr @nohr.se · 01/02/2026
Take me down to the Parallax city where the far moves slow and the near moves quickly
92166224819
Johan Edstedt @parskatt.bsky.social · 30/01/2026
Is there any current 3/4D reconstruction method able to accurately reconstruct this scene? (with correct size of thunderstorm)
3161
Johan Edstedt @parskatt.bsky.social · 26/01/2026
regardless of your views on AI, i strongly agree with this viewpoint! you are not funded by taxpayers to optimize your knowledge consumption workflows. it's good that you enjoy your job, you are getting paid to enjoy your job
050
Johan Edstedt @parskatt.bsky.social · 21/01/2026
061
Johan Edstedt @parskatt.bsky.social · 13/01/2026
Correspondence is a much prettier word than match. Pixel correspondence makes more sense than feature matching (what is a feature?).
3171
Johan Edstedt @parskatt.bsky.social · 12/01/2026
you're welcome
3180
Johan Edstedt @parskatt.bsky.social · 09/01/2026
Can someone please explain why 3DV papers are consistently interesting and ICCV/CVPR (3D) papers are consistently boring?
280
Johan Edstedt @parskatt.bsky.social · 07/01/2026
Something I predict for academic papers is that (complex) implementations/proofs will get significantly less credit than previously due to LLMs. Not sure if this will be a good or bad thing.
040
Johan Edstedt @parskatt.bsky.social · 19/12/2025
Have to say I like the "Nice ---" style of GPT models.
140
Johan Edstedt @parskatt.bsky.social · 17/12/2025
Sad to report that I now have the type of brain damage where you have to write 1024 instead of 1000.
2110
Johan Edstedt @parskatt.bsky.social · 15/12/2025
Since this take is so completely wrong I have to repost as well. The posted figure is only showing the top-50. No shit it's gonna show the biggest companies and unis. There's a productive conversation to be had, but this aint it. aiworld.eu/story/from-b...
0131
Johan Edstedt @parskatt.bsky.social · 11/12/2025
At this point I'm extremely allergic to any non top-level defaults.
040
Johan Edstedt @parskatt.bsky.social · 09/12/2025
CVPR submissions now have this retracted look, presumably due to previous OpenReview leaks.
530
Johan Edstedt @parskatt.bsky.social · 04/12/2025
Not sure if anyone noticed but there are pycolmap CUDA wheels since a month back: pypi.org/project/pyco...
pypi.org
pycolmap-cuda12
COLMAP bindings
1131
Johan Edstedt @parskatt.bsky.social · 01/12/2025
0171
Johan Edstedt @parskatt.bsky.social · 30/11/2025
My predictions: - debacle will not impact reviewing in a significant way - ratings will likely be similar next year - peer review will continue to struggle with review load
3191
Johan Edstedt @parskatt.bsky.social · 28/11/2025
Oof
2184
Johan Edstedt @parskatt.bsky.social · 27/11/2025
1180
Reposted by Johan Edstedt
Zhenjun Zhao @ericzzj.bsky.social · 25/11/2025
SPIDER: Spatial Image CorresponDence Estimator for Robust Calibration Zhimin Shao, Abhay Yadav, Rama Chellappa, Cheng Peng tl;dr: 3D VFM+2D ConvNet->feature extraction backbone; 3D descriptor head (for geometry)+2D warp head (for pattern) fusion arxiv.org/abs/2511.17750
041
Johan Edstedt @parskatt.bsky.social · 24/11/2025
Pixel reconstruction has recently been somewhat overshadowed by latent SSL approaches such as DINO. However, for 3D tasks we show that a scaled and simplified version of multi-view MAE (which we call MuM) can outperform DINOv3, all while using orders of magnitude less compute!
2141
Johan Edstedt @parskatt.bsky.social · 21/11/2025
1100
Johan Edstedt @parskatt.bsky.social · 21/11/2025
Put up my simple skysegmentation model on github over at github.com/Parskatt/sky... Results are pretty crisp, but it doesn't really deal with clouds (it's literally just a linear model on top of some coarse segmentation output).
2150