Submission: DGCF6D/YCB-V

Download submission
Submission name
Submission time (UTC) Sept. 26, 2024, 7:47 a.m.
User jysir
Task Model-based 6D localization of seen objects
Dataset YCB-V
Description
Evaluation scores
AR:0.817
AR_MSPD:0.869
AR_MSSD:0.832
AR_VSD:0.751
average_time_per_image:-1.000

Method: DGCF6D

User jysir
Publication
Implementation
Training image modalities RGB-D
Test image modalities RGB-D
Description

We trained our model using PBR data, employing YOLOX for object detection and segmentation, and used default 3D models without ICP refinement. The architecture consisted of a single model for all objects within each dataset, combining object detection and pose regression.

Author: Jingyang Liu.

Computer specifications RTX3090