Method: ZebraPoseSAT-EffnetB4(DefaultDetections-2023)

User annamalaipraveen@gmail.com
Publication ZebraPose: Coarse to Fine Surface Encoding for 6DoF Object Pose Estimation, CVPR2022
Implementation https://github.com/suyz526/ZebraPose
Training image modalities RGB
Test image modalities RGB
Description

Submitted to: BOP Challenge 2023

Training data: real + provided PBR

Used 3D models: Reconstructed for T-LESS, default for other datasets

Notes:

Setting: One network per object was trained

2D Bounding Box: Default detections of 2023

Modifications to the original ZebraPose paper:

Added Symmetry-Aware Training (SAT). The network and loss functions are not changed. There will be a new ground truth for the sym. objects, details can be found in the Github Repository. Special thanks to Yongliang Lin for his contribution.

We replace the Resnet34 backbone with EffnetB4 in ZebraPose. (Only replace the backbone in the pose estimation part)

About the submission to segmentation challenge 2023: For every 2D bounding box provided by a 2D detector, we use ZebraPose network to infer the object visible mask and binary codes (as we did for object pose estimation). And we save 1) the confidence score from the 2D detector 2) as well as the predicted visible object mask into the json file for the segmentation evaluation.

List of contributors:

Praveen Annamalai Nathan, Sandeep Prudhvi Krishna Inuganti,Yongliang Lin, Yongzhi Su,Yu Zhang, Didier Stricker, Jason Rambach

Computer specifications Intel(R) Xeon(R) E-2146G CPU @ 3.50GHz, Nvidia RTX2080Ti

Public submissions

Date Submission name Dataset
2023-09-25 20:28 - LM-O
2023-09-25 20:34 - TUD-L
2023-09-25 20:37 - YCB-V
2023-09-25 20:49 - HB
2023-09-25 21:06 - ITODD
2023-09-25 21:55 - IC-BIN
2023-09-25 22:04 - T-LESS
2023-09-26 13:33 - LM-O
2023-09-26 14:21 - IC-BIN
2023-09-26 14:45 - TUD-L
2023-09-26 16:36 - T-LESS
2023-09-26 16:39 - HB
2023-09-26 16:53 - YCB-V
2023-09-26 18:00 - ITODD