Submission: RACE6D_RGB/ITODD

Download submission
Submission name
Submission time (UTC) May 12, 2026, 4:47 p.m.
User helose
Task Model-based 6D localization of seen objects
Dataset ITODD
Description
Evaluation scores
AR:0.302
AR_MSPD:0.407
AR_MSSD:0.290
AR_VSD:0.209
average_time_per_image:0.013

Method: RACE6D_RGB

User helose
Publication RACE-6D: Real-time Accurate Coarse-to-finE object 6D Pose Transformer, CVPR 2026 Findings
Implementation Pytorch, code can be found at https://github.com/Yoonwoo-Ha/RACE-6D
Training image modalities RGB
Test image modalities RGB
Description

Training data: real + provided PBR

Used 3D models: Default for other datasets

Authors: Yoonwoo Ha, Hyungpil Moon (SungKyunKwan University).

For LMO, HB, ICBIN datasets, we only use the provided synthetic training data (PBR) in training. While for YCBV, TUDL, TLESS, we use the provided real data and synthetic data (PBR) in training.

For detection, we developed a unified pose estimation model that encompasses the object detection process

Computer specifications GPU RTX 3090; CPU intel i9-12900K