2026-09-18T20:59:18Z https://drops.dagstuhl.de/oai

oai:drops-oai.dagstuhl.de:25919 2026-09-05T19:32:07Z ddc:004 open_access

Model-Agnostic Uncertainty-Aware Semantic Segmentation with Conformal Risk Guarantees for Scene Understanding Badjie, Bakary Cecílio, José Friedrich, Nils-Jonathan Seyffer, Norman Jäger, Georg Casimiro, António semantic segmentation uncertainty quantification evidential deep learning conformal prediction risk control selective prediction Accurate and reliable scene segmentation is a fundamental requirement for autonomous navigation systems operating in open and dynamic environments. As these systems increasingly rely on data-driven perception modules, their safety and operational robustness hinge on well-calibrated uncertainty estimates that can support explicit control of prediction errors through conformal calibration. Most existing uncertainty-aware segmentation approaches remain architecture-specific and are not evaluated under a common uncertainty-and-calibration protocol across distinct segmentation architectures and datasets. This work introduces a model-agnostic conformal segmentation pipeline that enables operationally meaningful, calibration-based error control in real-world deployments. The proposed framework treats segmentation networks as black boxes and operates on per-pixel class probabilities that are fine-tuned through evidential deep learning (EDL) to decompose aleatoric and epistemic uncertainties. We then apply pixel-wise, class-conditional split-conformal calibration to derive acceptance thresholds for user-defined target error rates. We instantiate the pipeline with DINOv2, Mask2Former, and SegFormer and evaluate it on a newly collected Lisbon street scene (LiSS) dataset; additional cross-dataset results on COCO, using a restricted set of safety-relevant classes, are reported in the appendix. Results show architecture- and class-dependent in-domain uncertainty-error alignment and indicate that dataset shift weakens uncertainty-based filtering and conformal risk control. This motivates continuous monitoring and recalibration as a practical requirement for trustworthy segmentation in safety-critical navigation. Schloss Dagstuhl – Leibniz-Zentrum für Informatik Bakary Badjie and José Cecílio and Nils-Jonathan Friedrich and Norman Seyffer and Georg Jäger and António Casimiro 2026 Is Part Of OASIcs, Volume 143, 30th Ada-Europe International Conference on Reliable Software Technologies (AEiC 2026) InProceedings Text doc-type:ResearchArticle publishedVersion application/pdf doi:10.4230/OASIcs.AEiC.2026.1 urn:nbn:de:0030-drops-259199 https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.AEiC.2026.1 eng https://creativecommons.org/licenses/by/4.0/legalcode