SPK: Eliciting Structured Prior Knowledge for Interpretable Out-of-Distribution Detection in Real-Time Object Detection

Hugging Face Daily Papers Papers

Summary

Structured Prior Knowledge (SPK) is a framework that explicitly extracts latent semantic, geometric, and contextual priors from pretrained object detectors to achieve state-of-the-art out-of-distribution detection, improving interpretability and reliability.

Object detectors often produce over-confident predictions for objects outside their training categories, leading to so-called out-of-distribution (OoD) hallucinations. Existing approaches for detecting or mitigating such hallucinations typically either construct scoring functions directly over learned object detector representations or modify the object detector itself to suppress hallucination emergence. However, the latent priors implicitly encoded in these representations remain largely unexplored and have not been explicitly decoded for OoD detection. To uncover and exploit these latent priors, we propose Structured Prior Knowledge (SPK), a hallucination-oriented framework that explicitly elicits OoD-relevant priors from pretrained object detectors. Specifically, SPK leverages in-distribution data and hallucination-inducing samples as diagnostic supervision to elicit part-level semantic concepts underlying object detector decision-making, rather than using them merely for rejection or object detector adaptation. The elicited semantic priors are further integrated with geometric and contextual priors to form a compact five-dimensional SPK representation for OoD detection. Extensive experiments across diverse object detector architectures and multiple OoD benchmarks demonstrate that SPK achieves state-of-the-art OoD detection. Our findings reveal that pretrained object detectors already encode substantially richer latent knowledge than is typically exploited for OoD detection. More importantly, this knowledge can be explicitly elicited and organized into a compact, structured, and interpretable knowledge space for prediction reliability analysis. This suggests a promising proactive route for improving object detector reliability by explicitly uncovering and leveraging latent priors. Code and data are available at: https://gricad-gitlab.univ-grenoble-alpes.fr/dnn-safety/spk
Original Article
View Cached Full Text

Cached at: 08/20/26, 08:05 PM

Paper page - SPK: Eliciting Structured Prior Knowledge for Interpretable Out-of-Distribution Detection in Real-Time Object Detection

Source: https://huggingface.co/papers/2608.19080

Abstract

Structured Prior Knowledge explicitly extracts latent semantic, geometric, and contextual priors from pretrained object detectors to detect out-of-distribution hallucinations via a compact interpretable representation.

Object detectorsoften produce over-confident predictions for objects outside their training categories, leading to so-called out-of-distribution (OoD)hallucinations. Existing approaches for detecting or mitigating suchhallucinationstypically either construct scoring functions directly over learned object detector representations or modify the object detector itself to suppress hallucination emergence. However, thelatent priorsimplicitly encoded in these representations remain largely unexplored and have not been explicitly decoded for OoD detection. To uncover and exploit theselatent priors, we proposeStructured Prior Knowledge(SPK), a hallucination-oriented framework that explicitly elicits OoD-relevant priors from pretrainedobject detectors. Specifically, SPK leverages in-distribution data and hallucination-inducing samples as diagnostic supervision to elicitpart-level semantic conceptsunderlying object detector decision-making, rather than using them merely for rejection or object detector adaptation. The elicited semantic priors are further integrated with geometric andcontextual priorsto form a compact five-dimensional SPK representation for OoD detection. Extensive experiments across diverse object detector architectures and multiple OoD benchmarks demonstrate that SPK achieves state-of-the-art OoD detection. Our findings reveal that pretrainedobject detectorsalready encode substantially richer latent knowledge than is typically exploited for OoD detection. More importantly, this knowledge can be explicitly elicited and organized into a compact, structured, and interpretable knowledge space for prediction reliability analysis. This suggests a promising proactive route for improving object detector reliability by explicitly uncovering and leveraginglatent priors. Code and data are available at: https://gricad-gitlab.univ-grenoble-alpes.fr/dnn-safety/spk

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2608\.19080

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2608.19080 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.19080 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2608.19080 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Sparse Autoencoders for Interpretable Out-of-Distribution Detection

arXiv cs.LG

This paper introduces a novel method using sparse autoencoders (SAEs) to learn interpretable features from intermediate network activations for out-of-distribution (OOD) detection, achieving state-of-the-art performance and providing insights into how distribution shifts affect learned representations.

OPRD: On-Policy Representation Distillation

Hugging Face Daily Papers

OPRD proposes a new knowledge distillation method that aligns student and teacher hidden states across layers during on-policy rollouts, eliminating sampling variance from token-space KL estimation. Empirically, OPRD outperforms output-space baselines on math reasoning benchmarks (AIME 2024/2025, AIMO) while being 1.44x faster and using 54% less memory.