Real-Time Monocular Segmentation and Pose Tracking of Multiple Objects (ECCV'16)

Показать описание

The video shows a real-time system capable of segmenting multiple 3D objects and tracking their pose using a single RGB camera, based on prior shape knowledge. The system uses twist-coordinates for pose parametrization and a pixel-wise second-order optimization approach resulting in very robust tracking, especially in cases of fast motion and scale changes. The implementation runs at about 50–100 Hz on a commodity laptop when tracking a single object without relying on GPGPU computations.

Рекомендации по теме

Комментарии

I say "sorta" because, by default, it only works for a single image, but the modification to support a video file or a webcam is not overly difficult. It also doesn't seem to include the pose DETECTION mentioned in the 2018 paper, only pose TRACKING. So you'll have to either code pose detection yourself or figure out a way to manually initialize the first pose.

cmstudent

That would be nice, if the open source code is shared! Thanks for the amazing work!

altamediacompany

Unbelievable achievement. Congratulations!

junpengzhang

This is pretty cool! Don't know what it does, but it's pretty cool anyway!!!

NightmareBlade

Nice work! Would also like to know when will the code be available...

fionafan

Nice work . Will the code be avaliable?

Dr.tech-

this is really nice! how did you do the wireframe object? did u design it yourself or the computer generated it after detected the object?

elenix

Real-Time Monocular Segmentation and Pose Tracking of Multiple Objects (ECCV'16)

Real-Time Monocular Segmentation and Pose Tracking of Multiple Objects (ECCV'16)

ECCV'16 Demo: Real-Time Monocular Segmentation and Pose Tracking of Multiple Objects

2021 Real-time monocular recovery of 3D body pose and shape in the wild

Real-Time Monocular Pose Estimation of 3D Objects using Temporally Consistent Local Color Histograms

ICCV'17 Demo: Real-Time Monocular Pose Estimation of 3D Objects using TCLC-Histograms

Real-Time Monocular 6DoF Tracking of Textureless Objects Using Photometrically-Enhanced Edges

Monocular Real-time Hand Shape and Motion Capture using Multi-modal Data - CVPR 2020

Realtime monocular depth on Google Colab - OSSDC VisionAI

Real-time Monocular 3D Tracking on CPU

Real-time Monocular 3D Tracking on CPU

Monocular real-time 3D tracking on CPU

Monocular 3D Human Pose Estimation In The Wild Using Improved CNN Supervision - 3DV2017

Real Time Monocular Depth Estimation CVPR 2018

Real-Time Underwater 3D Mapping from a Monocular Camera

Elon Musk says losers use LiDAR. [Explanation video]

Real-time monocular image-based 6-DoF localization (Part 1)

AdasWorks Demonstration of Real-time Object Segmentation

Using a Single RGB Frame for Real Time 3D Hand Pose Estimation in the Wild (IEEE WACV 2018)

Robust Scale Estimation in Real-Time Monocular SFM for Autonomous Driving

Real-Time Hierarchical Scene Segmentation and Classification

G2L-Net: Global to Local Network for Real-Time 6D Pose Estimation With Embedding Vector Features

Research at NVIDIA: Deep Object Pose Estimation for Semantic Robotic Grasping of Household Objects

Real-Time 3D Segmentation of Cluttered Scenes for Robot Grasping

497 - Two-hand Global 3D Pose Estimation Using Monocular RGB