Seunghyeok Back1, Sangbeom Lee2, Kangmin Kim2, Joosoon Lee2, Sungho Shin3, Jemo Maeng2, and Kyoobin Lee2†
1Korea Institute of Machinery & Materials (KIMM)
2Gwangju Institute of Science and Technology (GIST)
3Hyundai Motors Company (Robotics Lab)
Accurate and efficient segmentation of unknown objects in unstructured environments is essential for robotic manipulation. Unknown Object Instance Segmentation (UOIS), which aims to identify all objects in unknown categories and backgrounds, has become a key capability for various robotic tasks. However, existing methods struggle with over-segmentation and under-segmentation, leading to failures in manipulation tasks such as grasping. To address these challenges, we propose QuBER (Quadruple Boundary Error Refinement), a novel error-informed refinement approach for high-quality UOIS. QuBER first estimates quadruple boundary errors—true positive, true negative, false positive, and false negative pixels—at the instance boundaries of the initial segmentation. It then refines the segmentation using an error-guided fusion mechanism, effectively correcting both fine-grained and instance-level segmentation errors. Extensive evaluations on three public benchmarks demonstrate that QuBER outperforms state-of-the-art methods and consistently improves various UOIS methods while maintaining a fast inference time of less than 0.1 seconds. Furthermore, we show that QuBER improves the success rate of grasping target objects in cluttered environments.
Quadruple Boundary Error Refinement (QuBER) is a novel error-informed refinement method that addresses both fine-grained and instance-level errors in unknown object instance segmentation (UOIS) models. It aims to improve the robustness of unknown object segmentation and grasping. QuBER predicts the quadruple boundary error, a pixel-wise representation of true positive (TP), true negative (TN), false positive (FP), and false negative (FN) errors in the instance boundaries of the initial segmentation. This error estimation provides specific refinement guidance, identifying regions to be modified. The Error Guidance Fusion (EGF) module integrates these error estimates into the refinement process, leading to improved segmentation accuracy.
@misc{back2024highqualityunknownobjectinstance,
title={High-quality Unknown Object Instance Segmentation via Quadruple Boundary Error Refinement},
author={Seunghyeok Back and Sangbeom Lee and Kangmin Kim and Joosoon Lee and Sungho Shin and Jemo Maeng and Kyoobin Lee},
year={2024},
eprint={2306.16132},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2306.16132},
}