Open Problems

Out-of-Distribution & Anomaly Detection

Representation Adaptation and End-to-End Learning for Anomaly Detection Beyond ImageNet Backbones

Barrier to removePartly addressed
Possible candidate · 3/5 runs4 papers report this50% from 2025+

Generated automatically from the limitations stated in 4 papers (CVPR, ICLR, ECCV), listed under Evidence. It is not a paper, and it does not come from papers submitted to CSPaper.

The problem

Modern visual anomaly and out-of-distribution (OOD) detection methods rely almost entirely on frozen feature representations extracted from ImageNet-pretrained backbones. In specialized domains such as medical imaging, industrial inspection, or scientific imaging, ImageNet features fail to encode domain-specific normal variation, capping downstream anomaly scoring accuracy. Because current pipelines do not adapt or learn backbone representations end-to-end from in-distribution data, anomaly detection is structurally unviable wherever high-quality pre-trained foundation models do not already exist.

Why it matters

Enables high-performance anomaly and OOD detection in non-natural and data-scarce domains where off-the-shelf vision foundation models lack representative semantic features.

Ways to approach it

Prior-work checks are free with an account. Results someone already ran are shown to everyone.

  1. 1

    Benchmark parameter-efficient adaptation (e.g., LoRA, adapter layers) on pre-trained backbones using normal-only self-supervised objectives (such as masked feature reconstruction or self-distillation), measuring anomaly AUROC across non-natural image datasets (e.g., histopathology, semiconductor inspection).

  2. 2

    Design an end-to-end representation learning objective that regularizes encoder updates against feature collapse while training exclusively on normal target data, measuring linear probe separation and anomaly detection precision/recall relative to frozen backbone baselines.

Have a different approach?

Describe how you would tackle this problem and we'll look for papers that already do it. Free; your text stays private.

Free · 3 checks per day

Why it might fail

If normal-only backbone fine-tuning inherently suffers representation collapse across non-contrastive objectives, or if scaled domain-specific foundation models quickly emerge that render local backbone adaptation unnecessary.

Evidence

Each paper's own statement of the limitation, verbatim.

Nearest existing work

Related open problems

Out-of-Distribution & Anomaly Detection

Scope to testPartly addressed

Cross-Modality Robustness and Evaluation of Anomaly Detection Methods Across Tabular and Unstructured Data

Anomaly and out-of-distribution (OOD) detection algorithms are developed in modality-specific silos, tested either exclusively on tabular benchmarks or on high-dimensional unstructured data relying on pre-trained vision/language foundation models. Because no unified empirical evaluation assesses these algorithms across both tabular and unstructured modalities, practitioners cannot determine whether tabular anomaly detectors fail on raw high-dimensional inputs or whether embedding-dependent detectors degrade when pre-trained feature extractors are absent. As a consequence, safety-critical domains with heterogeneous or non-standard data types lack validated guidance on algorithm selection.

Possible candidate · 2/5 runs3 papers report this33% from 2025+
Generated automatically, not curated by hand. Automated prior-work checks catch about a third of existing work, so treat this problem as a lead to investigate.