Pose-graph via Adaptive Image Re-ordering

Daniel Barath, Jana Noskova, Iván Eichhardt, Jiri Matas

tl;dr: Integrating pose-estimation-RANSAC with pose-graph queue of global SfM brings huge speed improvements.

#BMVC2022

https://bmvc2022.mpi-inf.mpg.de/0127.pdf

#computervision #deeplearning

#dmytrotweetsaboutDL

Thanks to everyone who I met during my poster session yesterday at #BMVC #bmvc2022 for the valuable feedback and ideas as to how I can extend and improve my research - on my way to the workshops today and excited to hear a talk by @dimadamen 🤘

Went to an interesting talk by phillip isola at #bmvc2022 yesterday which mentioned treating generative models as data++. Now I'm waiting to see the first sota result on imagenet training only on samples generated by a text2image diffusion model.

Excited to share our recent work on exploiting patches relations in self-supervised vision transformer (RelViT), by @guglielmocamporese Elena Izzo and myself (#VIMP group @UniPadova)

PAPER: https://arxiv.org/abs/2206.00481
CODE: Cooming soon!

We will present our work #BMVC2022