LibreCV

Computer Vision and Machine Learning community

[LibreCV @cv_libre]

state of the art #cvpr2021 for image matching using vision #transformers by @kornia_foss

#computervision #opensource #machinelearning #AI

[Kornia @kornia_foss]

Amazing LoFTR is in #kornia master branch now! 🥳
Implement state-of-the are image matching methods with 2 lines of code
@JiamingSuen ❤️

#CVPR2021 #opensource #computervision #DeepLearning

[Akshay 🚀 @akshay_pachaar]

Kornia is a Differentiable Library based on PyTorch.

💫 Build a 𝑭𝒂𝒄𝒆 𝑫𝒆𝒕𝒆𝒄𝒕𝒐𝒓 in just a few lines of Python code using Kornia.

A Thread 🧵 👇
@kornia_foss

[Nima Boscarino @NimaBoscarino]

Do you want to contribute to an open-source project? 🤔 Kornia, the popular Computer Vision library, is hosting a Docs ‘n Demos Sprint to showcase their algorithms – and they want your help! Feel free to ask for help in the GitHub issue or message me directly 🤗

[Kornia @kornia_foss]

What do you think about potential kornia_nonfree library, similar to OpenCV non_free?
It would cover some algorithms, released under restrictive licenses, such as R2D2, DISK, SuperPoint or others.

[Kornia @kornia_foss]

Yesterday we added SOLD² line segment matcher into #kornia.
Today we merged homography from line correspondences, as a first use-case for it :)

https://t.co/jewN9chijf

[Kornia @kornia_foss]

We have recently merged edge-aware blur PR, thanks to
nitaifingerhut
https://t.co/ZQe0vSuvVK
#imageprocessing #computervision #PyTorch #kornia

[Kornia @kornia_foss]

🚀Kornia 0.6.6 has been released !!

🔥 ParametrizedLine API, load_image support for @apple @windows, integration demos with @huggingface and many more.

📚Release notes:
https://t.co/a9bZMibC9b

#computervision #machinelearning #ai #deeplearning

[Kornia @kornia_foss]

Hybrid Skip: A Biologically Inspired Skip Connection for the UNet Architecture

Nikolaos Zioulis, Georgios Albanis, Petros Drakoulis, Federico Alvarez, Dimitrios Zarpalas, Petros Daras

tl;dr: new version of UNet architecture with help of #kornia

https://t.co/TO7OGXLGzX

[PyTorch @PyTorch]

Announcing PyTorch 1.12: TorchArrow, functorch, accelerated PyTorch training on Mac and more. Take a look: 👇
https://t.co/aPh68avBLt

[Dmytro Mishkin @ducha_aiki]

Levente Hajder gives a basic course about affine correspondences in computer vision. Simple and elegant geometry, which often not taught at computer vision courses nowadays.

at @CVPR
#CVPR2022

[Dmytro Mishkin @ducha_aiki]

Slides are here
https://t.co/OW6pUVmdfi

[Kornia @kornia_foss]

Landslide4Sense competition focuses on landslide detection using globally distributed multi-source satellite imagery.

👉 https://t.co/KZMyTT7GDT

kudos to @robisee and team to leverage #kornia augmentations for segmentation tasks.

#computervision #deeplearning #remotesensing

[Edgar @edgarriba]

Community call for @Apple #MacBookPro users !! Help us to improve @kornia_foss @PyTorch

[JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱 @JFPuget]

I'm using nnaudio + @kornia_foss for an audio classification project (not a @kaggle comp), and I get a 8x speedup compared to using librosa + albumentation.

Speedup is mostly due to moving data preparation from cpu to gpu.

[Kornia @kornia_foss]

🚀Kornia 0.6.5 has been released !!

🔥Image i/o module written in @rustlang, diamond_square and plasma augmentations, new geometric metrics and more.

📚Release notes:
https://t.co/WioUVgu57M

#computervision #machinelearning #ai #deeplearning

[Kornia @kornia_foss]

VQGAN-CLIP: Open Domain Image Generation and Editing with Natural Language Guidance

@RiversHaveWings, @BlancheMinerva, @jbusted1,@dashstander @EricHallahan @lcastricato
@EdwardRaffML

https://t.co/mXi0EHDtP0

tl;dr: in title :)
#kornia used for augmentation

[Kornia @kornia_foss]

Off-the-shelf LoFTR from #kornia with tuned image size (840 by large size) + tuned RANSAC threshold = 0.723 o public leaderboard, top-10 so far.
#IMC2022 #kaggle
https://t.co/intsZs6NiL

[Dmytro Mishkin @ducha_aiki]

Examples & tutorials.
We provide 3 inference (untuned) baselines, with accuracy scores:
1) OpenCV SIFT + MNN + MAGSAC++: 0.165
2) @kornia_foss KeyNet + AffNet + HardNet8 + SMNN + MAGSAC++: 0.543
3) @jatentaki DISK + MNN + MAGSAC+: 0.502
Good luck!
3/3

[Dmytro Mishkin @ducha_aiki]

Datasets: The val/test dataset is similar to GoogleUrban (last year).
Training data: IMC Phototourism, but feel free to use other datasets as well.
2/

[FOSSASIA Summit @fossasiasg]

Jian Shi, Core #Maintainer @kornia_foss will give a presentation on 'Accelerate Your Data Augmentation Pipeline with Kornia'. Check it out https://t.co/Ds3HtuTYQg