🤖 AI Summary
This study systematically evaluates the robustness and generalization of instance segmentation models under realistic image degradations (15 types of noise, blur, etc.) and cross-domain data (diverse acquisition settings). Leveraging benchmarks including COCO, we establish a multi-dimensional evaluation framework to comparatively analyze mainstream architectures, normalization strategies (Group Normalization vs. Batch Normalization), pretraining paradigms, and single- versus multi-stage detection frameworks. Key findings include: (i) Group Normalization significantly enhances degradation robustness (+8.2% mAP), whereas Batch Normalization improves cross-domain generalization (+5.7% mAP); (ii) multi-stage detectors exhibit scale-invariant performance, while single-stage models suffer from poor resolution generalization. We release a reproducible robustness leaderboard and an evidence-based model selection guide, providing critical insights for deploying instance segmentation models in real-world scenarios and informing architectural design and pretraining strategy choices.
📝 Abstract
This article presents a comprehensive evaluation of instance segmentation models with respect to real-world image corruptions as well as out-of-domain image collections, e.g., images captured by a different set-up than the training dataset. The out-of-domain image evaluation shows the generalization capability of models, an essential aspect of real-world applications, and an extensively studied topic of domain adaptation. These presented robustness and generalization evaluations are important when designing instance segmentation models for real-world applications and picking an off-the-shelf pretrained model to directly use for the task at hand. Specifically, this benchmark study includes state-of-the-art network architectures, network backbones, normalization layers, models trained starting from scratch versus pretrained networks, and the effect of multitask training on robustness and generalization. Through this study, we gain several insights. For example, we find that group normalization (GN) enhances the robustness of networks across corruptions where the image contents stay the same but corruptions are added on top. On the other hand, batch normalization (BN) improves the generalization of the models across different datasets where statistics of image features change. We also find that single-stage detectors do not generalize well to larger image resolutions than their training size. On the other hand, multistage detectors can easily be used on images of different sizes. We hope that our comprehensive study will motivate the development of more robust and reliable instance segmentation models.
