FocalPose++：レンダリングと比較による焦点距離とオブジェクト姿勢の推定
FocalPose++は、既知のオブジェクトの単一RGB入力画像からカメラとオブジェクトの6D姿勢とカメラの焦点距離を共同で推定する、レンダリングと比較に基づく新しい手法であり、従来手法よりも低いエラー率を実現します。
GS2Pose: A Two-Stage Approach for 6D Object Pose Estimation Using Gaussian Splatting Without CAD Models
GS2Pose is a novel two-stage method for estimating the 6D pose of novel objects from RGB-D images, leveraging 3D Gaussian Splatting (3DGS) to achieve accuracy and robustness without relying on CAD models.
Modular Quantization-Aware Training (MQAT) for Efficient 6D Object Pose Estimation in Resource-Constrained Environments
MQAT, a novel quantization-aware training method, leverages the modular structure of 6D object pose estimation networks to achieve significant model compression while maintaining or even improving accuracy, making it ideal for resource-constrained applications.
Object Pose Estimation of Transparent Objects Using Neural Radiance Fields for Render-and-Compare
This research paper introduces a novel method for estimating the 6D pose of transparent objects from a single RGB image by integrating Neural Radiance Fields (NeRF) into a render-and-compare pipeline, demonstrating superior performance compared to traditional methods relying on textured meshes, particularly for challenging transparent and reflective objects.
Addressing Pose Ambiguity in 6D Object Pose Estimation via Score-Based Diffusion on SE(3)
A novel score-based diffusion method applied to the SE(3) group, marking the first application of diffusion models to SE(3) within the image domain, specifically tailored for pose estimation tasks.
© 2024 by Linnk AI