RSCS6D: Keypoint Extraction-Based 6D Pose Estimation

In this work, we propose an improved network, RSCS6D, for 6D pose estimation from RGB-D images by extracting keypoint-based point clouds. Our key insight is that keypoint cloud can reduce data redundancy in 3D point clouds and accelerate the convergence of convolutional neural networks. First, we em...

Full description

Saved in:
Bibliographic Details
Main Authors: Weiyu Liu, Nan Di
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Applied Sciences
Subjects:
Online Access:https://www.mdpi.com/2076-3417/15/12/6729
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850157002072784896
author Weiyu Liu
Nan Di
author_facet Weiyu Liu
Nan Di
author_sort Weiyu Liu
collection DOAJ
description In this work, we propose an improved network, RSCS6D, for 6D pose estimation from RGB-D images by extracting keypoint-based point clouds. Our key insight is that keypoint cloud can reduce data redundancy in 3D point clouds and accelerate the convergence of convolutional neural networks. First, we employ a semantic segmentation network on the RGB image to obtain mask images containing positional information and per-pixel labels. Next, we introduce a novel keypoint cloud extraction algorithm that combines RGB and depth images to detect 2D keypoints and convert them into 3D keypoints. Specifically, we convert the RGB image to grayscale and use the Sobel edge detection operator to identify 2D edge keypoints. Additionally, we compute the Curvature matrix from the depth image and apply the Sobel operator to extract keypoints critical for 6D pose estimation. Finally, the extracted 3D keypoint cloud is fed into the 6D pose estimation network to predict both translation and rotation.
format Article
id doaj-art-cfdfae5f58464eafb6b4b4e23c07e4ac
institution OA Journals
issn 2076-3417
language English
publishDate 2025-06-01
publisher MDPI AG
record_format Article
series Applied Sciences
spelling doaj-art-cfdfae5f58464eafb6b4b4e23c07e4ac2025-08-20T02:24:18ZengMDPI AGApplied Sciences2076-34172025-06-011512672910.3390/app15126729RSCS6D: Keypoint Extraction-Based 6D Pose EstimationWeiyu Liu0Nan Di1Academy for Advanced Interdisciplinary Studies, Northeast Normal University, No. 5268 Renmin Avenue, Changchun 130024, ChinaAcademy for Advanced Interdisciplinary Studies, Northeast Normal University, No. 5268 Renmin Avenue, Changchun 130024, ChinaIn this work, we propose an improved network, RSCS6D, for 6D pose estimation from RGB-D images by extracting keypoint-based point clouds. Our key insight is that keypoint cloud can reduce data redundancy in 3D point clouds and accelerate the convergence of convolutional neural networks. First, we employ a semantic segmentation network on the RGB image to obtain mask images containing positional information and per-pixel labels. Next, we introduce a novel keypoint cloud extraction algorithm that combines RGB and depth images to detect 2D keypoints and convert them into 3D keypoints. Specifically, we convert the RGB image to grayscale and use the Sobel edge detection operator to identify 2D edge keypoints. Additionally, we compute the Curvature matrix from the depth image and apply the Sobel operator to extract keypoints critical for 6D pose estimation. Finally, the extracted 3D keypoint cloud is fed into the 6D pose estimation network to predict both translation and rotation.https://www.mdpi.com/2076-3417/15/12/67296D pose estimationkeypoint extractionconvolutional neural networkSobel operator
spellingShingle Weiyu Liu
Nan Di
RSCS6D: Keypoint Extraction-Based 6D Pose Estimation
Applied Sciences
6D pose estimation
keypoint extraction
convolutional neural network
Sobel operator
title RSCS6D: Keypoint Extraction-Based 6D Pose Estimation
title_full RSCS6D: Keypoint Extraction-Based 6D Pose Estimation
title_fullStr RSCS6D: Keypoint Extraction-Based 6D Pose Estimation
title_full_unstemmed RSCS6D: Keypoint Extraction-Based 6D Pose Estimation
title_short RSCS6D: Keypoint Extraction-Based 6D Pose Estimation
title_sort rscs6d keypoint extraction based 6d pose estimation
topic 6D pose estimation
keypoint extraction
convolutional neural network
Sobel operator
url https://www.mdpi.com/2076-3417/15/12/6729
work_keys_str_mv AT weiyuliu rscs6dkeypointextractionbased6dposeestimation
AT nandi rscs6dkeypointextractionbased6dposeestimation