CenterMask3D: Are Bounding-Box Based Approaches Suitable for Volumetric Instance Segmentation?

Leipert M, Zabler S, Maier A (2026)

Publication Type: Conference contribution

Publication year: 2026

Book Volume: 31 (3)

Conference Proceedings Title: Special Issue of e-Journal of Nondestructive Testing

Event location: Linz AT

URI: https://www.ndt.net/search/docs.php3?id=32619

DOI: 10.58286/32619

Open Access Link: https://www.ndt.net/article/ctc2026/papers/ict26_Contribution_253.pdf

Abstract

We evaluate CenterMask3D, an anchor-free volumetric instance segmentation method adapted from FCOS3D, for the analysis of industrial Computed Tomography data. The study utilizes a specialized dataset of CT scans featuring boxed shoes, characterized by complex geometries and overlapping instance bounding boxes. To address significant memory constraints and limited training samples, we implement a memory-efficient Residual Squeeze-and-Excitation backbone and an extensive on-thefly data augmentation pipeline. Additionally we improve a lightweight Mask-Head based on channel-wise attention. Our model achieves a mean Intersection over Union (mIoU) per instance of 0.365 and an AP@IoU=0.5 of 0.317. While the results demonstrate that CenterMask3D successfully segments individual volumetric objects, the performance highlights inherent limitations of bounding-box-based approaches in dense volumetric settings. Specifically, the accumulation of errors from noisy box proposals and interpolation artifacts suggests that kernel-based alternatives, such as PanopticFCN or Flood Filling Networks, may offer more robust pathways for high-fidelity volumetric instance segmentation.

Authors with CRIS profile

Martin Leipert Lehrstuhl für Informatik 5 (Mustererkennung) Andreas Maier Lehrstuhl für Informatik 5 (Mustererkennung)

Involved external institutions

Technische Hochschule Deggendorf / Deggendorf Institute of Technology DE Germany (DE)

How to cite

APA:

Leipert, M., Zabler, S., & Maier, A. (2026). CenterMask3D: Are Bounding-Box Based Approaches Suitable for Volumetric Instance Segmentation? In ndt.net (Eds.), Special Issue of e-Journal of Nondestructive Testing. Linz, AT.

MLA:

Leipert, Martin, Simon Zabler, and Andreas Maier. "CenterMask3D: Are Bounding-Box Based Approaches Suitable for Volumetric Instance Segmentation?" Proceedings of the 15th Conference on Industrial Computed Tomography (iCT), Linz Ed. ndt.net, 2026.

BibTeX: Download