- Rating
- 3
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
The paper proposes HIVE-AD, a training-free, few-shot industrial anomaly detection (IAD) framework built on a frozen DINOv2 backbone. The method addresses three design axes that the authors argue are underserved by prior training-free pipelines: (i) representation depth — Variance-Guided Layer Weighting (VGLW) computes a spatial-variance statistic over the normal support features of the deepest selected layer and deterministically maps it to one of three fixed layer-weight templates (early/balanced/late emphasis) over five selected DINOv2 blocks {11, 17, 23, 29, 39}; (ii) output granularity — Decoupled Pixel-Ranking Scoring (DPRS) separates image-level ranking (Gaussian smoothing + top-ratio pooling) from pixel-level scoring; (iii) spatial extent — Pixel-Level Soft Spatial Aggregation (PSSA) blends a smoothed anomaly map with a local max-filtered response to consolidate fragmented responses while remaining score-continuous. Patch anomaly evidence comes from layer-wise normal memory banks with exact cosine 1-NN matching. An optional frozen MLLM (InternVL2-8B) converts localized regions into structured defect reports, but is explicitly decoupled from all quantitative predictions. Validation covers MVTec-AD and VisA under 1/2/4-shot settings. In the 1-shot setting, HIVE-AD reports 97.8/99.0 I-AUROC/I-AUPR, 97.6 P-AUROC, 94.6 PRO on MVTec-AD, and 94.5/95.3, 98.2, 93.4 on VisA. It improves image-level AUROC over the bare DINOv2-G baseline by 2.3 and 2.8 points, ablations show each component contributes, and a sensitivity analysis (Table 3) shows VGLW is stable across threshold choices and a continuous interpolation. The supplement provides a single global configuration (Table A2), runtime/memory measurements, category-level results, a decision-threshold diagnostic (ZS vs. oracle accuracy), an automatic audit of the report branch (location/text-align/type accuracy), and a limitations discussion.
Critical strengths
- Rigorous and transparent evaluation protocol with a genuinely global configuration. All results (main paper + supplement) use one fixed hyperparameter set (Table A2) shared across datasets, categories, shot counts, and backbone scales, with no category-conditioned overrides. The VGLW sensitivity study (Table 3) directly probes the most heuristic component and shows I-AUROC varies by at most 0.45/0.54 points across six settings. The authors explicitly mark VisA PRO as a loss to SubspaceAD and explicitly state (B.7) that the comparison table is not a compute-matched backbone replacement study. This level of calibration transparency is above average for this subfield and materially strengthens the credibility of the headline numbers.
- Clean architectural separation between quantitative detection and language reporting. The MLLM branch is invoked only after scores, decisions, and maps are finalized, its output is neither fused into nor allowed to veto the visual result (B.5), and enabling/disabling it leaves all metrics unchanged. The automatic audit (Table A8) reports modest numbers (e.g., location 62.9, type 65.3 on MVTec-AD 1-shot) and the authors themselves caution that these are "diagnostic rather than evidence of fully faithful explanations." This honest treatment avoids the common failure of overclaiming MLLM-based anomaly explanation.
Critical weaknesses
- Limited conceptual novelty; each component is a straightforward heuristic, and the gains over the closest training-free baselines are modest. VGLW is a 3-way hard thresholding of hand-fixed weight templates; DPRS is Gaussian smoothing plus top-ratio pooling; PSSA is a max-filter blend. The 1-shot improvement over SubspaceAD is +0.7 I-AUROC on MVTec-AD and +1.3 on VisA (Sec. 4.2), while on VisA PRO HIVE-AD actually loses to SubspaceAD (93.4 vs. 95.5 in the 1-shot row of Table 1), and at 4-shot on VisA it also trails in I-AUROC (95.3 vs. 94.7 is a gain, but P-AUROC 98.6 vs. 98.4 and PRO 94.7 vs. 96.0 again show mixed results). The ablation (Table 2) shows the components are complementary, but the underlying mechanism of why the spatial-variance statistic should map monotonically to depth emphasis is only asserted (Sec. 3.2), not analyzed. For a paper whose core claim is a new way of organizing evidence, the conceptual depth behind the mapping is thin.
- Provenance of the global hyperparameters is unclear, which undermines the "no tuning" claim. The paper emphasizes that (τ_1,τ_2)=(0.8,1.5), (σ_r,r)=(4.0,0.002), (σ_p,h,α)=(8.0,31,0.6) form "one global configuration shared by both datasets" (Sec. 4.1). But it does not state how these values were chosen — in particular whether they were selected using the same test sets on which they are reported. Table 3 shows the default (0.8, 1.5) achieves "best or tied-best on four of six dataset-metric combinations," which reads as if the default was verified (or selected) against test performance. The same concern applies to the support-set choice for K-shot evaluation (no variance over random support draws is reported anywhere, despite single-support results like macaroni2 1-shot I-AUROC of 84.8 vs. 81.6 at 4-shot in Table A6 suggesting high support sensitivity). Additionally, the decision-threshold diagnostic (Table A3) reveals a large gap between zero-shot accuracy (82.9 / 73.2) and oracle accuracy (95.7 / 88.7) at 1-shot, indicating the CutPaste-style threshold calibration is far from effective — the binary-decision pathway is weak even though the paper correctly excludes it from the main claims.
Questions for authors
- Tables 1/A4/A5/A6 report point estimates. Could you report mean ± std over multiple random draws of the K-shot support set (or at least the minimum/maximum over draws) for the headline numbers? The large category-level swings (e.g., macaroni2, fryum) suggest support-set sensitivity may be significant at K = 1.
- In Eq. (3), v_s is computed without ℓ2 normalization from raw frozen features, yet the matching in Eq. (2) uses ℓ2-normalized features. What is the empirical scale/distribution of v_s across the 27 categories of MVTec-AD + VisA? Does the three-way template assignment actually vary across categories, or does nearly every category land in one bucket (which would reduce VGLW to a fixed weighting)?
- Sec. H attributes the VisA PRO gap to "under-coverage of large or fragmented anomalous regions." A simple diagnostic (e.g., PRO restricted to images with large GT-connected components, or coverage vs. PRO curves for varying α) would substantiate this claim. Could you provide it?
- Since DPRS already decouples ranking from localization, extending the localization branch with a lightweight multi-scale PSSA (as you suggest in Sec. H) seems the most direct route to closing the PRO gap; I would encourage this for a future version rather than requiring it now.
4 comments
Response to Reviewer 8jHo (Q1) (Part II)
· Category-level results on MVTec-AD and VisA in the 1-shot, 2-shot, and 4-shot settings
MVTec-AD
| Category | 1-shot I-AUROC | 1-shot P-AUROC | 2-shot I-AUROC | 2-shot P-AUROC | 4-shot I-AUROC | 4-shot P-AUROC |
|---|---|---|---|---|---|---|
| bottle | 99.94 | 98.97 | 99.94 | 99.00 | 100.00 | 99.07 |
| cable | 94.65 | 96.59 | 95.47 | 97.06 | 98.19 | 97.56 |
| capsule | 97.48 | 98.15 | 95.58 | 98.29 | 96.85 | 98.52 |
| carpet | 99.94 | 99.30 | 99.98 | 99.31 | 100.00 | 99.34 |
| grid | 99.99 | 99.40 | 99.99 | 99.42 | 100.00 | 99.47 |
| hazelnut | 99.00 | 99.40 | 99.59 | 99.46 | 100.00 | 99.52 |
| leather | 99.99 | 99.25 | 99.99 | 99.25 | 100.00 | 99.22 |
| metal_nut | 99.92 | 96.29 | 99.99 | 97.10 | 100.00 | 97.55 |
| pill | 96.76 | 95.02 | 96.36 | 95.44 | 97.25 | 95.87 |
| screw | 81.43 | 97.90 | 77.28 | 97.63 | 84.09 | 98.29 |
| tile | 100.00 | 97.56 | 99.99 | 97.77 | 100.00 | 97.74 |
| toothbrush | 99.91 | 98.62 | 99.80 | 98.86 | 100.00 | 98.98 |
| transistor | 96.89 | 93.19 | 97.74 | 94.91 | 98.39 | 95.65 |
| wood | 99.56 | 96.71 | 99.42 | 96.99 | 99.88 | 96.86 |
| zipper | 99.89 | 97.80 | 99.88 | 97.71 | 100.00 | 97.86 |
| Mean | 97.69 ± 0.07 | 97.61 ± 0.06 | 97.40 ± 0.18 | 97.88 ± 0.03 | 98.31 ± 0.43 | 98.10 ± 0.17 |
VisA
| Category | 1-shot I-AUROC | 1-shot P-AUROC | 2-shot I-AUROC | 2-shot P-AUROC | 4-shot I-AUROC | 4-shot P-AUROC |
|---|---|---|---|---|---|---|
| candle | 93.55 | 99.39 | 94.32 | 99.46 | 94.32 | 99.49 |
| capsules | 96.14 | 98.24 | 96.76 | 98.80 | 97.75 | 99.04 |
| cashew | 95.68 | 97.44 | 96.29 | 97.95 | 96.93 | 98.15 |
| chewinggum | 99.25 | 99.65 | 99.51 | 99.63 | 99.85 | 99.60 |
| fryum | 96.20 | 93.86 | 95.81 | 94.41 | 96.35 | 94.76 |
| macaroni1 | 90.16 | 98.80 | 92.86 | 99.03 | 92.45 | 99.13 |
| macaroni2 | 86.29 | 98.35 | 84.97 | 98.70 | 86.36 | 99.31 |
| pcb1 | 94.88 | 99.16 | 93.39 | 99.23 | 93.60 | 99.28 |
| pcb2 | 91.59 | 98.32 | 92.28 | 98.80 | 93.59 | 98.97 |
| pcb3 | 94.37 | 98.44 | 96.83 | 98.82 | 97.92 | 99.01 |
| pcb4 | 98.70 | 97.68 | 98.99 | 97.96 | 99.68 | 98.16 |
| pipe_fryum | 98.75 | 98.83 | 99.55 | 98.97 | 99.36 | 99.02 |
| Mean | 94.63 ± 0.30 | 98.18 ± 0.13 | 95.13 ± 0.55 | 98.48 ± 0.07 | 95.68 ± 0.46 | 98.66 ± 0.03 |
Response to Reviewer 8jHo (Q2)
Q2.
· Different normalization procedures in Eqs. (3) and (2)
We thank the reviewer for drawing attention to the different normalization procedures used for spatial variance and cosine matching. We compute in Eq. (3) before applying the additional patch-wise normalization. This statistic measures the variance of the normal support features from the deepest selected layer across spatial positions and then averages it over support images and channels. It describes spatial variation within the normal reference images. The normalization in Eq. (2) serves cosine-distance matching, so the anomaly score depends on differences in feature direction. The two computations provide the signal for layer-weight selection and the normalized features for matching, respectively.
We further compare the category rankings before and after normalization across three support-set draws. The rank correlation between the category means of the original and those of the variance after patch-wise normalization is on MVTec-AD and on VisA. This additional diagnostic shows that the relative category rankings are highly consistent under the two statistics, providing further evidence for selecting layer weights based on spatial variation in normal support features.
· Empirical scale and threshold intervals of
We thank the reviewer for asking about the numerical scale of and its relationship to the templates. We use the same pair of thresholds, 0.8 and 1.5, throughout. The early template corresponds to , the uniform template to , and the late template to . Among the categories examined across three draws, cashew, carpet, grid, leather, tile, and wood consistently fall in the middle interval, and chewinggum and macaroni2 consistently fall in the upper interval. For screw, falls in the upper interval in all three draws, and the selected late template achieves the highest I-AUROC among the candidates in each comparison. These empirical intervals directly link the observed template assignments to the scale of the statistic. The revised manuscript will report and the template assignment for each category using the same procedure, so readers can see how categories from both datasets are distributed across the threshold intervals.
· Variation in template assignment across categories and support sets
We thank the reviewer for suggesting that we check whether the three templates actually vary across categories. In the three draws used for this assignment diagnostic, several categories consistently select uniform, while chewinggum and macaroni2 consistently select late. These assignments illustrate differences in weighting preferences across categories. Once the support features determine the template, its weights remain fixed for all query images associated with that support set. This preserves adaptability to the support set in few-shot settings and ensures consistent evaluation conditions within each support set.
Response to Reviewer 8jHo (Q3)
Q3.
· Diagnostic setup for examining insufficient region coverage on VisA
We thank the reviewer for suggesting a coverage diagnostic based on anomaly region shape. On VisA, we use three random seeds to draw support sets. For each seed, we keep the support set, query set, and other parameters fixed and vary the PSSA mixing weight α over 0, 0.2, 0.4, 0.6, and 0.8. We report coverage for large regions, regions with multiple connected components, and regions with a single connected component based on the size and connectivity of the ground-truth regions. The multi-component group is used to examine fragmented anomalies. We compute all coverage values at 5% FPR and report the mean across the three random seeds. A fixed false-positive operating point allows direct comparison of coverage at different values of α and shows which types of anomaly regions benefit.
| α | Large-Region Coverage (%) | Multi-Component Region Coverage (%) | Single-Component Region Coverage (%) |
|---|---|---|---|
| 0 | 88.26 | 85.15 | 94.65 |
| 0.2 | 89.52 | 86.47 | 94.90 |
| 0.4 | 90.22 | 87.10 | 94.86 |
| 0.6 | 90.54 | 87.23 | 94.54 |
| 0.8 | 90.55 | 86.96 | 94.06 |
At α = 0, coverage for large and multi-component regions is 88.26% and 85.15%, respectively, which is 6.39 and 9.50 percentage points lower than coverage for single-component regions. As α increases from 0 to 0.6, coverage improves steadily for both groups. At the default setting of α = 0.6, coverage increases by 2.28 percentage points for large regions and by 2.08 percentage points for multi-component regions, while single-component coverage decreases by only 0.11 percentage points. These results show that the main benefit of PSSA is to reinforce the responses in large and fragmented anomaly regions, improving coverage for these groups at the same false positive rate. The results for the individual groups support the explanation in Section H that region coverage may affect PRO on VisA.
Response to Reviewer 8jHo (Q1) (Part I)
We thank the reviewer for the encouraging assessment, for recognizing our work on a unified evaluation configuration, experimental transparency, and the decoupling of visual detection from language reporting, and for raising questions about variation across support-set draws, spatial variance and template assignment, and anomaly region coverage on VisA.
Q1.
· Mean and standard deviation of overall metrics across random support sets
We thank the reviewer for suggesting that we report results across multiple random support sets. For each shot setting on both datasets, we use three random seeds to draw the required number of support images from the normal training images in each category. For each evaluation, we first compute the category macro-average and then report the mean ± standard deviation across the three runs, as shown below.
MVTec-AD
VisA
In the 1-shot setting, I-AUROC is 97.69 ± 0.07 on MVTec-AD and 94.63 ± 0.30 on VisA. Across the two datasets and three shot settings, the standard deviation ranges from 0.07 to 0.55 percentage points for I-AUROC and from 0.03 to 0.17 percentage points for P-AUROC. The largest standard deviation is 0.72 percentage points for 2-shot PRO on MVTec-AD. On both datasets, mean P-AUROC and PRO increase with the number of support images per category, suggesting that more normal reference images help the visual branch achieve better localization in repeated evaluations.