Submission: GDRNPPDet_PBRReal/HB

Download submission
Submission name
Submission time (UTC) Oct. 14, 2022, 12:23 p.m.
User zyMeteroid
Task Model-based 2D detection of seen objects
Dataset HB
Training model type None
Training image type Synthetic (only PBR images provided for BOP Challenge 2020 were used)
Description
Evaluation scores
AP:0.809
AP50:0.958
AP75:0.900
AP_large:0.911
AP_medium:0.827
AP_small:0.363
AR1:0.828
AR10:0.845
AR100:0.850
AR_large:0.953
AR_medium:0.854
AR_small:0.488
average_time_per_image:0.081

Method: GDRNPPDet_PBRReal

User zyMeteroid
Publication Not yet
Implementation Pytorch, code can be found at https://github.com/shanice-l/gdrnpp_bop2022
Training image modalities RGB
Test image modalities RGB
Description

GDRNPPDet for BOP2022

Authors: Xingyu Liu, Ruida Zhang, Chenyangguang Zhang, Bowen Fu, Jiwen Tang, Xiquan Liang, Jingyi Tang, Xiaotian Cheng, Yukang Zhang, Gu Wang, and Xiangyang Ji (Tsinghua University).

In the PBRReal setting, for LMO, HB, ICBIN and ITODD 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. We trained one model for each dataset.

GDRNPPDet was based on YOLOX. We used stronger data augmentation and ranger optimizer.

Computer specifications GPU RTX 3090; CPU AMD EPYC 7H12 64-Core Processor.