[CVPR 2020] SuperGlue: Learning Feature Matching with Graph Neural Networks (5min oral)

Автор: Paul-Edouard Sarlin

Загружено: 2020-07-25

Просмотров: 10445

Описание:

This is the 5 minute oral video for our CVPR 2020 paper:
"SuperGlue: Learning Feature Matching with Graph Neural Networks"
Project Page: https://psarlin.com/superglue
Paper: https://arxiv.org/abs/1911.11763
Code: https://github.com/magicleap/SuperGlu...
Authors: Paul-Edouard Sarlin, Daniel DeTone, Tomasz Malisiewicz, Andrew Rabinovich

Abstract:
This paper introduces SuperGlue, a neural network that matches two sets of local features by jointly finding correspondences and rejecting non-matchable points. Assignments are estimated by solving a differentiable optimal transport problem, whose costs are predicted by a graph neural network. We introduce a flexible context aggregation mechanism based on attention, enabling SuperGlue to reason about the underlying 3D scene and feature assignments jointly. Compared to traditional, hand-designed heuristics, our technique learns priors over geometric transformations and regularities of the 3D world through end-to-end training from image pairs. SuperGlue outperforms other learned approaches and achieves state-of-the-art results on the task of pose estimation in challenging real-world indoor and outdoor environments. The proposed method performs matching in real-time on a modern GPU and can be readily integrated into modern SfM or SLAM systems.

[CVPR 2020] SuperGlue: Learning Feature Matching with Graph Neural Networks (5min oral)

Доступные форматы для скачивания:

Скачать видео mp4

Информация по загрузке:

Скачать аудио mp3

Похожие видео

SuperGlue: Learning Feature Matching with Graph Neural Network

SuperGlue: Learning Feature Matching with Graph Neural Network

An Introduction to Graph Neural Networks: Models and Applications

An Introduction to Graph Neural Networks: Models and Applications

[ECCV 2024] GeoCalib: Learning Single-image Calibration with Geometric Optimization

[ECCV 2024] GeoCalib: Learning Single-image Calibration with Geometric Optimization

[CVPR 2023] OrienterNet: Visual Localization in 2D Public Maps with Neural Matching

[CVPR 2023] OrienterNet: Visual Localization in 2D Public Maps with Neural Matching

Deep Visual SLAM Frontends: SuperPoint, SuperGlue, and SuperMaps (#CVPR2020 Invited Talk)

Deep Visual SLAM Frontends: SuperPoint, SuperGlue, and SuperMaps (#CVPR2020 Invited Talk)

Jon Barron - Understanding and Extending Neural Radiance Fields

Jon Barron - Understanding and Extending Neural Radiance Fields

Deep learning on graphs: successes, challenges | Graph Neural Networks | Michael Bronstein

Deep learning on graphs: successes, challenges | Graph Neural Networks | Michael Bronstein

LoFTR 5-minutes Introduction Video (CVPR 2021)

LoFTR 5-minutes Introduction Video (CVPR 2021)

Управление поведением LLM без тонкой настройки

Управление поведением LLM без тонкой настройки

System Design Concepts Course and Interview Prep

System Design Concepts Course and Interview Prep

SuperGlue: Learning Feature Matching With Graph Neural Networks

SuperGlue: Learning Feature Matching With Graph Neural Networks

Самая сложная модель из тех, что мы реально понимаем

Самая сложная модель из тех, что мы реально понимаем

LSTM is dead. Long Live Transformers!

LSTM is dead. Long Live Transformers!

[ICCV 2021] Pixel-Perfect Structure-from-Motion with Featuremetric Refinement

[ICCV 2021] Pixel-Perfect Structure-from-Motion with Featuremetric Refinement

ICLR 2021 Keynote -

ICLR 2021 Keynote - "Geometric Deep Learning: The Erlangen Programme of ML" - M Bronstein

[NeurIPS 2023] SNAP: Self-Supervised Neural Maps for Visual Positioning and Semantic Understanding

[NeurIPS 2023] SNAP: Self-Supervised Neural Maps for Visual Positioning and Semantic Understanding

Learning 3D Reconstruction in Function Space

Learning 3D Reconstruction in Function Space

«Эгоистичный ген» Ричарда Докинза [Veritasium]

«Эгоистичный ген» Ричарда Докинза [Veritasium]

SIGGRAPH 2022: Adversarial Skill Embeddings

SIGGRAPH 2022: Adversarial Skill Embeddings

[CVPR 2021] Back to the Feature: Learning Robust Camera Localization from Pixels to Pose

[CVPR 2021] Back to the Feature: Learning Robust Camera Localization from Pixels to Pose