Trapped in Texture Bias? A Large Scale Comparison of Deep Instance Segmentation

J. Theodoridis, Jessica Hofmann, J. Maucher, A. Schilling
2/14/2026

No official code found. Best implementation: JohannesTheo/trapped-in-texture-bias (16 stars, MIT, last commit 2024-01-16).

Abstract

Do deep learning models for instance segmentation generalize to novel objects in a systematic way? For classification, such behavior has been questioned. In this study, we aim to understand if certain design decisions such as framework, architecture or pre-training contribute to the semantic understanding of instance segmentation. To answer this question, we consider a special case of robustness and compare pre-trained models on a challenging benchmark for object-centric, out-of-distribution texture. We do not introduce another method in this work. Instead, we take a step back and evaluate a broad range of existing literature. This includes Cascade and Mask R-CNN, Swin Transformer, BMask, YOLACT(++), DETR, BCNet, SOTR and SOLOv2. We find that YOLACT++, SOTR and SOLOv2 are significantly more robust to out-of-distribution texture than other frameworks. In addition, we show that deeper and dynamic architectures improve robustness whereas training schedules, data augmentation and pre-training have only a minor impact. In summary we evaluate 68 models on 61 versions of MS COCO for a total of 4148 evaluations.

View on arXivView PDF

Code Implementations(1)

JohannesTheo/trapped-in-texture-biasCommunity100%
MIT

Official code release for the paper Trapped in texture bias? A large scale comparison of deep instance segmentation, accepted at ECCV 2022

163Python, Jupyter NotebookJul 16, 20222 years agoMIT
classificationcomputer-visiondeepl-learningdetectioneccv2022+4 more

Similar Papers

PartDistillation: Learning Parts from Instance Segmentation

85% match2026

PerSense: Training-Free Personalized Instance Segmentation in Dense Images

83% match1 repo312024

CLIMB-3D: Continual Learning for Imbalanced 3D Instance Segmentation

83% match1 repo32025

Self-Balanced R-CNN for Instance Segmentation

83% match1 repo62024

ISBNet: a 3D Point Cloud Instance Segmentation Network with Instance-aware Sampling and Box-aware Dynamic Convolution

83% match1 repo1692026

MaskUno: Switch-Split Block For Enhancing Instance Segmentation

83% match2024

Cite this paper

@article{theodoridis2026trapped,
  title  = {Trapped in Texture Bias? A Large Scale Comparison of Deep Instance Segmentation},
  author = {J. Theodoridis and Jessica Hofmann and J. Maucher and A. Schilling},
  year   = {2026},
  eprint = {2401.09109},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/2401.09109},
  journal = {ECCV 2024 2024}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2401.09109.svg)](https://codeofpaper.com/paper/2401.09109)

Discussion