We propose a feature-fusion network for pose estimation directly from RGB images without any depth infor-mation in this study.First,we introduce a two-stream architecture consisting of segmentation and regression streams.The segmentation stream processes the spatial embedding features and obtains the corresponding image crop.These features are further coupled with the image crop in the fusion network.Second,we use an efficient perspective-n-point(E-PnP)algorithm in the regression stream to extract robust spatial features between 3D and 2D keypoints.Finally,we perform iterative refinement with an end-to-end mechanism to improve the estimation performance.We conduct experiments on two public datasets of YCB-Video and the challenging Occluded-LineMOD.The results show that our method outperforms state-of-the-art approaches in both the speed and the accuracy.

Xiao-Long Yang;Xiao-Hong Jia;Yuan Liang;Lu-Bin Fan

Academy of Mathematics and Systems Science,Chinese Academy of Sciences,Beijing 100190,China;University of Chinese Academy of Sciences,Beijing 100049,China;Alibaba DAMO Academy,Alibaba Group,Hangzhou 311121,China

计算机科学技术学报（英文版）

[1]Xiao-Long Yang;Xiao-Hong Jia;Yuan Liang;Lu-Bin Fan-.6D Object Pose Estimation in Cluttered Scenes from RGB Images)[J].计算机科学技术学报（英文版）,2022(03):719-730

Cluttered,keypoints,YCB,LineMOD

6D,Object,Pose,Estimation,Scenes,from,RGB,Images,We,propose,fusion,network,estimation,directly,images,without,any,depth,infor,this,study,First,introduce,architecture,consisting,segmentation,regression,streams,processes,spatial,embedding,features,obtains,corresponding,crop,These,are,further,coupled,Second,use,efficient,perspective,PnP,algorithm,extract,robust,between,2D,Finally,iterative,refinement,end,mechanism,improve,performance,conduct,experiments,public,datasets,Video,challenging,Occluded,results,show,that,our,method,outperforms,state,art,approaches,both,speed,accuracy

0.59533