Noisy Annotations in Semantic Segmentation
This work addresses annotation quality issues in segmentation tasks, which is crucial for researchers and practitioners relying on accurate labels, but it is incremental as it builds on existing benchmarks and noise simulation approaches.
The paper tackles the problem of noisy annotations in semantic segmentation by simulating noise conditions and introducing new benchmarks (COCO-N, Cityscapes-N, COCO-WAN) to assess model robustness and generalization, revealing limitations in existing methods for handling label noise.
Obtaining accurate labels for instance segmentation is particularly challenging due to the complex nature of the task. Each image necessitates multiple annotations, encompassing not only the object class but also its precise spatial boundaries. These requirements elevate the likelihood of errors and inconsistencies in both manual and automated annotation processes. By simulating different noise conditions, we provide a realistic scenario for assessing the robustness and generalization capabilities of instance segmentation models in different segmentation tasks, introducing COCO-N and Cityscapes-N. We also propose a benchmark for weakly annotation noise, dubbed COCO-WAN, which utilizes foundation models and weak annotations to simulate semi-automated annotation tools and their noisy labels. This study sheds light on the quality of segmentation masks produced by various models and challenges the efficacy of popular methods designed to address learning with label noise.