Machine Vision and Applications · 2025
Naturally Constrained Reject Option Classification
+1.3% coverageWe present the invited journal extension of our reject-option work, showing how per-class binomial thresholds translate to dependable abstentions on ImageNet, long-tailed wildlife data, and remote-sensing pipelines. The paper details how to productionize the method with dataset-specific coverage targets, selective accuracy dashboards, and policy deployment checklists for ML and analyst teams.
Highlights
- Invited journal extension of the ISVC Best Paper that introduces deployment-minded calibration tooling.
- Per-class binomial thresholds outperform global temperature scaling on ImageNet and remote sensing splits.
Artifacts & reproduction
PyTorch toolkit for per-class reject-option training with binomial threshold search, dashboards, and CLI.
- Tune per-class thresholds on ImageNet, iNat, or custom datasets with one command.
- Export coverage/accuracy curves and selective accuracy plots for reports.
- Integrate abstention policies into existing Torch models via lightweight hooks.
- Production-ready toolkit that underpins selective prediction guardrails across multiple domains (vision, text, and 2-D data).