SeNeDiF-OOD: Semantic Nested Dichotomy Fusion for Out-of-Distribution Detection Methodology in Open-World Classification. A Case Study on Monument Style Classification

TL;DR

SeNeDiF-OOD introduces a semantic nested dichotomy fusion framework for open-world OOD detection in image classification, replacing single-score decisions with a cascade of semantically grounded binary gates. The method combines hierarchical decision fusion with semantic priors and heterogeneous expert detectors, aiming for robust filtering while preserving in-distribution accuracy. The authors validate the approach through a detailed case study on MonuMAI monument style classification, showing substantial gains in precision and AUROC and offering layer-wise diagnostics that reveal how different OOD modalities are addressed. This work advances interpretable, robust, and actionable OOD handling for real-world open-world vision systems, with practical implications for cultural heritage applications and beyond.

Abstract

Out-of-distribution (OOD) detection is a fundamental requirement for the reliable deployment of artificial intelligence applications in open-world environments. However, addressing the heterogeneous nature of OOD data, ranging from low-level corruption to semantic shifts, remains a complex challenge that single-stage detectors often fail to resolve. To address this issue, we propose SeNeDiF-OOD, a novel methodology based on Semantic Nested Dichotomy Fusion. This framework decomposes the detection task into a hierarchical structure of binary fusion nodes, where each layer is designed to integrate decision boundaries aligned with specific levels of semantic abstraction. To validate the proposed framework, we present a comprehensive case study using MonuMAI, a real-world architectural style recognition system exposed to an open environment. This application faces a diverse range of inputs, including non-monument images, unknown architectural styles, and adversarial attacks, making it an ideal testbed for our proposal. Through extensive experimental evaluation in this domain, results demonstrate that our hierarchical fusion methodology significantly outperforms traditional baselines, effectively filtering these diverse OOD categories while preserving in-distribution performance.

Figures (10)

• Figure 1: Examples of MonuMAI misclassifying OOD inputs as architectural elements.
• Figure 2: Nested set view of the data regime: the far world surrounds the near world, followed by progressively closer superclasses until reaching the problem’s known world. The figure motivates layered OOD handling: coarse filters are effective at the periphery, whereas near the known world, decisions must be semantics-aware and carefully calibrated.
• Figure 3: Schematic of the proposed cascade with nested dichotomies. Each layer enforces a semantically aligned gate. Samples flow top-down: at every stage, out-of-scope inputs are rejected and the remainder becomes progressively more specific, from open-world data to the task’s known world (ID).
• Figure 4: Schematic of the SeNeDiF-OOD architecture instantiated for the MonuMAI use case.
• Figure 5: Layer-wise performance heatmap across the main metrics (Accuracy, Precision, Recall, F1, AUROC). Each row corresponds to a gate in the SeNeDiF-OOD pipeline (far-world filter; building vs. no-building; monument vs. building; known vs. unknown style).