CountZES: Counting via Zero-Shot Exemplar Selection

Muhammad Haris Khan; Muhammad Ibraheem Siddiqui

REVIEW 4 major objections 4 minor 5 references

CountZES: Counting via Zero-Shot Exemplar Selection

T0 review · 4 major / 4 minor · reviewed 2026-08-03 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read CountZES claims that zero-shot object counting can be done without any task-specific training by selecting three complementary exemplars—semantic, density-guided, and feature-consensus—from fixed pretrained models.

desk verdict Genuinely useful inference-only exemplar-selection pipeline for zero-shot counting with strong cross-domain results, but the abstract overstates 'superior performance' against its own Table 1 and the density-guided stage is self-referential. read the letter →

arxiv 2512.16415 v3 pith:YUXQOBRB submitted 2025-12-18 cs.CV

Muhammad Ibraheem Siddiqui , Muhammad Haris Khan This is my paper

classification cs.CV

keywords zero-shotobjectcountingexemplarselectioninference-onlyopen-vocabularydetectiondensityestimationpointpromptingcross-domaingeneralizationfoundationmodels

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that counting instances of a category never seen during training, given only its name, can be solved by choosing good exemplars rather than by training a counter. It introduces CountZES, an inference-only pipeline that selects one exemplar from each of three complementary stages: text-conditioned detection, density-map consistency, and feature-space consensus. These three exemplars are fed together to a frozen pretrained density estimator, and the authors report that this selection strategy outperforms other inference-only methods across natural, aerial, and medical images. The central claim is that exemplar quality—not regression capacity—is the bottleneck in zero-shot counting, and that a deliberately diverse set of three exemplars suffices. If true, counting a new class would require no annotations or fine-tuning, only a class name and a few frozen foundation models.

What carries the argument

The central object is the multi-stage exemplar selection pipeline, CountZES. The Detection-Anchored Exemplar (DAE) stage refines a detector box into a single-instance region by scoring SAM masks with text-alignment and entropy. The Density-Guided Exemplar (DGE) stage turns density-map peaks into point prompts for SAM, filters boxes by predicted region count, and selects the box whose count is closest to the mode of all candidate counts while remaining semantically compact. The Feature-Consensus Exemplar (FCE) stage clusters SAM feature embeddings of single-instance boxes and picks the one nearest the majority-cluster centroid. The mechanism is the deliberate diversity of the resulting three-

What would settle it

Take classes where the segmenter is known to split objects (e.g., watches segment into dial and band, eggs into yolk and white). If replacing the segmenter with one that yields complete masks does not reduce CountZES's error on those classes, then exemplar selection itself is not what is driving the reported accuracy; the same could be tested by measuring whether selected exemplar masks cover at least 80% of the annotated instance.

Watch

Extended reading notes

Core claim

On its own terms, the paper discovers that zero-shot object counting can be recast as an exemplar-inference problem governed by semantic, statistical, and feature-level priors. CountZES refines a coarse open-vocabulary detection into a single-instance exemplar using text-conditioned similarity peaks and a segmenter; it then derives a second exemplar from density-map peaks whose region count and semantic compactness agree with a pseudo-ground-truth mode; and it selects a third exemplar by clustering segmenter features and choosing the box nearest the majority-cluster centroid. The authors claim this one-exemplar-per-stage diversity is what drives accuracy, and they support the claim with expe

Load-bearing premise

The entire pipeline presumes that a single point prompt reliably makes the segmenter return the whole object, not a fragment of it; when it returns only a dial, a yolk, or a character on a book cover, the selected exemplar is partial and the final count is conditioned on that partial object.

Editorial extensions

If this is right

A user can count arbitrary categories—cars in aerial views, cells in microscopy—from a class name alone, with no exemplar annotation and no task-specific optimization.
Any improvement to a single component (detector, segmenter, density estimator) can be dropped into the pipeline directly, since all components remain fixed during inference.
Because text is used only as an inference-time probe, the method avoids the text-count entanglement that the paper identifies as a source of cross-domain brittleness in other zero-shot counters.
On the reported benchmarks, CountZES matches or beats several trained zero-shot counting methods while using only frozen pretrained components.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A direct corollary the paper leaves implicit is that per-class accuracy should be predictable from the segmenter's mask completeness for that class; this can be tested by correlating per-class MAE with manual mask-completeness scores.
The one-exemplar-per-stage rule is a fixed design choice; an adaptive weighting of stages based on detector confidence, density spread, or feature-cluster tightness could improve counts on images where one stage is unreliable.
The same exemplar-selection logic could extend to other dense prediction tasks, such as promptable instance segmentation or referring localization, where a text prompt must be converted into representative visual support without labels.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Quick take: CountZES is a real, useful entry in zero-shot object counting — a three-stage, inference-only exemplar-selection pipeline with strong cross-domain results — but the paper sells itself as better than its own Table 1, and the density-guided stage has a self-referential loop that deserves scrutiny.

What's actually new is the combination, not the parts. CLIP, GroundingDINO, SAM, density-peak sampling, and feature clustering are known; arranging them as Detection-Anchored, Density-Guided, and Feature-Consensus stages, each contributing one exemplar to a fixed density estimator, is a sensible and apparently effective scheme. The ablations show incremental gains (25.13 → 20.75 → 15.77 MAE), and the cross-domain numbers are the strongest part: on CARPK, CountZES gets 7.24 MAE against T2ICount's 8.61 and GeCo's 10.34; on the medical sets it beats both by large margins. The limitation section (A.7) honestly documents SAM fragmenting objects into partial masks, with figures. Credit where it's due.

Soft spots, roughly in order.

First, claim calibration. The abstract says "superior performance among ZOC methods" and the conclusion repeats it, but Table 1 shows in-domain FSC-147 MAE 15.77 against 11.76 for T2ICount and 13.30 for GeCo. The body text is more careful — "competitive MAE and lowest RMSE" is what the table actually supports. The stress-test charge about the cross-domain row (CountingDINO's 20.93 vs 21.09) mostly dissolves if you accept the asterisk exclusion: CountingDINO uses box prompts, so it isn't playing the same zero-shot game, and among genuinely prompt-only inference methods CountZES leads that row. But the exclusion needs to be argued, not just footnoted, and the abstract should not claim superiority over all ZOC methods when the standard benchmark says otherwise.

Second, DGE is partly self-referential. The same pretrained density estimator finds the density peaks, computes the RoI counts, produces the pseudo-GT as the KDE mode of those same counts, scores boxes by closeness to that pseudo-GT, and then receives the winning exemplar to produce the final count. Choosing an exemplar that agrees with the estimator's own prior is a reasonable heuristic, and the ablation shows it helps, but the paper should say plainly that the pseudo-GT is not an independent signal. As written, "count-driven consistency" reads stronger than the evidence supports.

Third, the SAM partial-mask problem is real but documented, and it is confined to the appendix. Given that A.7 is one of the better sections, the main text should point to it and note which classes fail. Fourth, minor: the abstract promises code with no link, and most hyperparameters (α, w_sim/w_ent, the RoI interval) get no sensitivity analysis beyond the k=16 ablation.

Who this is for: people working on class-agnostic or zero-shot counting. The engineering is solid, the cross-domain results are the headline, and the flaws are addressable. I would send it to review, with the authors asked to reconcile the abstract with Table 1 and add an honest paragraph on the DGE loop. A serious referee can get this into good shape.

Referee Report

4 major / 4 minor

Summary. CountZES proposes an inference-only pipeline for zero-shot object counting (ZOC) that discovers three exemplars per image through Detection-Anchored (DAE), Density-Guided (DGE), and Feature-Consensus (FCE) stages, then feeds them to a frozen pretrained density estimator. The method uses CLIP, GroundingDINO, SAM, and either CounTR or DSALVANet as fixed backbones, and claims superior ZOC performance while generalizing to natural, aerial, and medical domains. The paper reports results on FSC-147, CARPK, PerSense-D, VGG, and MBM, with ablations attributing gains to each stage.

Significance. If the headline claims held, the contribution would be a modular, training-free exemplar-selection strategy for ZOC, useful in settings where per-class fine-tuning is impractical. The paper is clearly written and includes sensible ablations, cross-domain protocols, and a runtime analysis. It also deserves credit for being explicit about its modularity and for reporting a limitations appendix. However, the central 'superior performance' claim is not supported by the paper's own main table on FSC-147, and the documented dependence on SAM's ability to produce complete single-instance masks imposes a stronger constraint than the abstract conveys. The significance of the work is moderate and conditional on a more careful framing of what is actually demonstrated.

major comments (4)

[Abstract, Sec. 1, Sec. 5, and Table 1] The claim of 'superior performance among ZOC methods' is contradicted by Table 1 on FSC-147. In the in-domain setting, CountZES reports MAE 15.77, while T2ICount, GeCo, and CountGD report 11.76, 13.30, and 14.76 respectively. The paper's own Section 4 text describes the in-domain MAE as 'competitive,' which is accurate, but the abstract and conclusion use 'superior' without qualification. Even in the cross-domain row, CountingDINO* has lower MAE (20.93 vs 21.09) and much lower RMSE (71.37 vs 110.14). If CountingDINO is excluded because it uses box prompts, that exclusion must be stated wherever 'superior' appears. As written, the paper is internally inconsistent about its main claim.
[Sec. 3.2, Eqs. (4), (5), (10)-(12)] The DGE stage selects an exemplar using the same density estimator that later produces the final count. Density peaks (Eq. 4), the RoI-count single-instance filter (Eq. 5), the per-box count estimates (Eq. 10), and the pseudo-GT KDE mode (Eqs. 11-12) all come from the same frozen DE. Consequently, the DGE gain in Table 5 may reflect that an exemplar is consistent with the counter's own prior rather than that it is a better exemplar in any transferable sense. This is a circularity risk for the claim that DGE improves exemplar quality. Please provide an analysis where the selection-time DE differs from the final-count DE, or where exemplars selected by DGE are evaluated with an independent counter, to separate self-consistency from true representativeness.
[Appendix A.7 and Figs. 13-14] The paper explicitly documents that SAM point prompts produce partial masks for watches, nail polishes, comic books, eggs, and sauce bottles, and states that 'residual noise can still propagate when the majority of candidate exemplars are imperfect.' This is a direct limitation on the central generalization claim, because all three stages rely on SAM masks as candidate boxes. Since several of these classes (e.g., eggs, comic books) appear in the benchmark classes used for evaluation, the unconditional wording of the abstract and conclusion is too strong. Please quantify how often partial-mask failures occur across the test splits and report per-class or per-infection performance, or restrict the claims to classes where SAM provides complete instances. A limitation appendix is welcome, but it must be reflected in the paper's central claims.
[Sec. 4, Tables 3 and 4] The cross-domain comparisons on PerSense-D and VGG/MBM include very few inference-only baselines: only TFOC on PerSense-D, and TFOC on VGG/MBM. Since the paper positions CountZES against inference-only methods, the absence of CountingDINO, TFCAC, OmniCount, and Count Anything from these tables makes it difficult to substantiate 'superior performance among inference-only methods' outside FSC-147. Please add the missing baselines or explicitly justify their exclusion.

minor comments (4)

[Table 3] The column header 'Density-based Class-wise' is confusing: the table appears to show density-based splits, not class-wise results. Also, the 'Overall' MAE for GeCo (13.12) is higher than each of its Low/Med/High MAE entries, which suggests an inconsistency in how the overall row is computed or labeled.
[Appendix A.3, Eq. (10)] Eq. (10) states \hat{c}_i \in \mathbb{Z}, but the density estimator returns a continuous count (and Eq. (5) uses fractional predictions like 1<\hat{c}<2). The codomain should be nonnegative reals.
[References] The bibliography contains Langley (2000), but I do not see a citation to it in the text. Please either cite it or remove it.
[Throughout] Minor formatting issues: 'Figure.' appears with inconsistent spacing, 'V A-Count' and 'DA VE' formatting is awkward, and Table 1 has two 'CountZES (Ours)' rows that could be distinguished more clearly with explicit 'in-domain' and 'cross-domain' labels in the rows.

Circularity Check

1 steps flagged · score 2.0 of 10

DGE's density-guided selection is self-referential (same density estimator supplies pseudo-GT and final count), but no prediction reduces to its inputs by construction and the central comparisons are external.

other [Section 3.2 (DGE), Eqs. (5), (11)-(12); final-count sentence in Section 3.3]
"To verify whether bi captures a single instance, we compute its integrated density within the region of interest (RoI): c(bi) = ... (5) ... we estimate a pseudo-GT count via consensus among all single-instance candidates ... cpseudo = arg max c p(c). (12) ... Ultimately, bDAE, bDGE, and bFCE constitute the final exemplar set, which is jointly provided as multiple support exemplars to the density estimator to produce the final density map and count prediction."

DGE selects its exemplar by closeness to a pseudo-GT count that is itself the mode of per-box counts computed with the same pretrained density estimator that later produces the final count. Thus the 'count consistency' term in J_DGE is a self-consistency loop: the exemplar is chosen to agree with the estimator's own statistics, and the final count is then generated by that same estimator. This is not a full reduction because the final count is not equal to the pseudo-GT by any equation, but it means the DGE stage does not provide independent evidence of exemplar quality or counting accuracy.

full rationale

The main derivation chain is not circular in the strong sense: CountZES uses fixed pretrained components (CLIP, GroundingDINO, SAM, a pretrained density estimator) and its final count comes from a density estimator applied to the selected exemplar set. No equation in the paper forces the final count to equal a fitted parameter or a pseudo-GT. The DGE stage, however, is self-referential: it uses the same density estimator both to define pseudo-GT (Eqs. 11-12) and to produce the final count, so its selected exemplar is chosen for self-consistency rather than external validation. This is a genuine methodological caveat, but the paper is transparent about calling it 'self-supervised.' The ablation (Table 5) does not exhibit Eq. X = Eq. Y by construction. The external comparisons on FSC-147, CARPK, PerSense-D, VGG, and MBM are benchmark-reported numbers, so the central empirical claim is falsifiable and not definitionally forced. Notably, the paper's own Table 1 shows CountZES is not the best on FSC-147 (MAE 15.77 vs T2ICount 11.76; cross-domain MAE 21.09 vs CountingDINO 20.93), which undermines the 'superior performance compared to SOTA' wording; that is a correctness/internal-consistency concern, not circularity. Appendix A.7 also explicitly documents SAM partial-mask failures, which limits the accuracy claim but again is not a circular-derivation issue. No load-bearing self-citation or imported uniqueness theorem is present; the only self-citation is the PerSense-D dataset definition, which is not an argumentative premise. Overall, the paper has one self-referential selection mechanism but no constructive reduction, so a modest circularity score is appropriate.

Assumptions & free parameters 8 free parameters · 7 assumptions · 0 invented entities

CountZES contributes a selection policy, not new representational machinery. It piggybacks on CLIP, GroundingDINO, SAM, off-the-shelf density estimators, and AnyUp upsampling, and it introduces several hand-set hyperparameters. No new physical entities or generative mechanisms are postulated. The central claim is therefore bounded by the quality and reliability of these pretrained components.

free parameters (8)

alpha in Eqs. (1) and (6) = 0.5
Balances detection/semantic confidence versus entropy-based purity in DAE and DGE scoring; fixed by hand with no sensitivity analysis.
w_sim / w_ent in Eq. (3) = 0.5 / 0.5
Weights for semantic strength versus spatial purity in SSES mask scoring; fixed without ablation.
SSES percentile start, step, and max peaks = p=90, Δp=10, k=16
Adaptive relaxation for peak detection; k is ablated on FSC-147, but percentile start and step are not.
Density peak threshold T_D = μ_D + 2σ_D
Heuristic to select high-confidence density peaks; chosen without distributional justification or sensitivity analysis.
RoI single-instance interval = (1, 2)
Boxes with integrated density between 1 and 2 are considered single-instance; interval set by fiat.
GroundingDINO detection threshold = 0.15
Detection confidence cutoff used for box proposals; not ablated.
FRES cluster count = 2
K-means with cosine similarity into majority/minority clusters; no ablation of the cluster count.
Entropy histogram bins and Laplace smoothing = B=20, ε=1e-8
Used for entropy estimates in DAE and DGE; fixed without sensitivity analysis.

assumptions (7)

domain assumption CLIP text-image similarity is a reliable dense semantic grounder for arbitrary class names.
Used throughout DAE to identify text-aligned regions and score masks; not validated for the specific dense scenes or classes.
domain assumption GroundingDINO returns at least one text-aligned detection box per image.
DAE's initial box selection and DGE's box gating depend on nonzero detections; no fallback is described for zero detections.
domain assumption SAM point prompts yield complete, single-instance masks.
All candidate exemplars are generated from SAM masks; Appendix A.7 and Figs. 13–14 document cases where masks are partial (watch dial, egg yolk, comic characters).
domain assumption Off-the-shelf density estimators (CounTR/DSALVANet) produce density maps whose peaks and RoI counts are reliable in the target domain.
P2P prompting, single-instance filtering, and pseudo-GT estimation all rely on the predicted density map; the limitations section notes cross-domain peak noise.
domain assumption The KDE mode of per-box count estimates is a valid pseudo-GT.
Eqs. (11)–(12) use KDE mode, with fallback to rounded mean; this assumes the per-box count distribution is informative and not systematically biased.
domain assumption Window-attention-based feature upsampling (AnyUp) preserves enough spatial detail for small boxes.
FRES upsamples SAM's 64×64 feature map to 256×256 using AnyUp; no quantitative validation of small-box feature fidelity is provided.
domain assumption Entropy and percentile-rank scores correlate with instance purity.
DAE and DGE use normalized entropy of similarity maps as a proxy for single-object coherence; this is a heuristic not derived from data.

how reviews work

0 comments

Cite this review

Pith. "Pith review of CountZES: Counting via Zero-Shot Exemplar Selection." pith.science (2026). https://pith.science/paper/YUXQOBRB

@misc{pith2026251216415,
  author       = {Pith},
  title        = {Pith review of: CountZES: Counting via Zero-Shot Exemplar Selection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YUXQOBRB}},
  note         = {Machine review of arXiv:2512.16415}
}

read the original abstract

Object counting in complex scenes is particularly challenging in the zero-shot (ZS) setting, where instances of unseen categories are counted using only a class name. Existing ZS counting methods that infer exemplars from text often rely on off-the-shelf open-vocabulary detectors (OVDs), which in dense scenes suffer from semantic noise, appearance variability, and multi-instance proposals. Alternatively, random image-patch sampling is employed, which fails to accurately delineate object instances. Since counting is sensitive to exemplar quality, such selection strategies often yield poorly representative exemplars, leading to inaccurate count estimation. To address these issues, we propose CountZES, an inference-only approach for object counting via ZS exemplar selection. CountZES discovers diverse exemplars through three synergistic stages: Detection-Anchored Exemplar (DAE), Density-Guided Exemplar (DGE), and Feature-Consensus Exemplar (FCE). DAE refines OVD detections to isolate precise single-instance exemplars. DGE introduces a density-driven, self-supervised paradigm to identify statistically consistent and semantically compact exemplars, while FCE reinforces visual coherence through feature-space clustering. Together, these stages yield a complementary exemplar set that balances textual grounding, count consistency, and feature representativeness. Experiments on diverse datasets demonstrate CountZES superior performance among ZOC methods while generalizing effectively across domains.

Figures

Figures reproduced from arXiv: 2512.16415 by the authors.

**Figure 1.** Zero-shot object counting comparison of our CountZES against T2ICount (Qian et al., 2025) , GeCo (Pelhan et al., 2024a), and TFOC (Shi et al., 2024) across diverse benchmarks spanning natural, aerial, and medical domains. retraining whenever a new class is introduced, limiting their scalability. To overcome this, class-agnostic counting has emerged, aiming to generalize to unseen categories (Yang et al., 2021; Huang… view at source ↗

**Figure 2.** (a) Detectors like GroundingDINO (Liu et al., 2024) often yield very few or even fail to provide single instance boxes. (b) Counting is sensitive to exemplar choice. (c) Introducing CountZES a multi-stage inference-only exemplar selection approach for ZOC. (Best viewed in zoom). et al., 2023) leverage image–text alignment to model object–class correlations without relying on visual exemplars, enhancing scalability… view at source ↗

**Figure 3.** Overview of CountZES. The pipeline operates in three stages. DAE refines GroundingDINO detections using CLIP similarity via SSES to obtain text-aligned single-instance regions. DGE then adopts a density-driven paradigm: it generates a density map conditioned on DAE exemplar, leverages P2P prompting to identify candidate exemplars, applies RoI-based single-instance filtering, and selects the most reliable exemplar th… view at source ↗

Figures from the paper (11 more)

**Figure 4.** Figure 4: Overview of GGES in the DGE stage. Per-box counts from single-instance candidates are used to estimate a pseudoGT count via density estimator. A SAM-based similarity map, conditioned on bDAE, captures semantic correspondence, and a composite score based on count proxi…

**Figure 5.** Figure 5: Overview of FRES module in the FCE stage. Singleinstance boxes Bsingle are projected onto SAM’s upsampled feature map Φ up to obtain ℓ2-normalized regional embeddings. These embeddings are clustered, and the box closest to the majoritycluster centroid is selected as …

**Figure 6.** Figure 6: Qualitative comparison of CountZES with T2ICount (Qian et al., 2025) and GeCo (Pelhan et al., 2024a) across diverse scenarios, including natural (FSC-147 and PerSense-D), aerial (CARPK), and medical cell-counting (VGG and MBM) images. Class: Marbles, GT: 51 Image DAE S…

**Figure 7.** Figure 7: Demonstrates complementary nature of DAE, DGE and FCE stages. For each stage, the top three exemplars are shown with their counting outputs. One exemplar from each stage (arrows) provides complementary cues, and their combination in CountZES improves overall counting p…

**Figure 8.** Figure 8: Visualization of SSES peak detection under multiinstance and single-instance coarse boxes. (Top row) When the initial detector box contains multiple objects, SSES identifies multiple high-similarity peaks within the coarse region, each serving as a point prompt to SA…

**Figure 9.** Figure 9: Illustration of the SSES module within the DAE stage. Given the detector-derived coarse box (in yellow), SSES leverages the text-conditioned CLIP similarity map to identify high-confidence response peaks using an adaptive percentile-based relaxation strategy. These pea…

**Figure 10.** Figure 10: Overview of P2P prompting and single-instance filtering. Density peaks extracted from the density map serve as point prompts to SAM, whose masks are filtered via RoI-count to retain only single-instance regions before being passed to the GGES module for final selectio…

**Figure 11.** Figure 11: Performance–runtime tradeoff among inference-only zero-shot counting methods, where MAE is plotted against average inference time (sec/image) and bubble size represents RMSE. CountZES achieves the best accuracy with competitive inference efficiency. A.4. Datasets FSC…

**Figure 12.** Figure 12: Additional qualitative results of CountZES on FSC-147 dataset. Input Image Density Map (DAE Exemplar) P2P Prompting Candidate Exemplars CountZES Class: Watches, GT: 111 Pred: 104 [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]

**Figure 13.** Figure 13: Failure case illustrating CountZES selection of incomplete exemplar boxes for the class “watches.” Due to SAM’s limited semantic awareness, point prompts from density peaks segment only the watch dial, excluding the wristband, which produces partial masks and incomple…

**Figure 14.** Figure 14: Additional failure cases highlighting CountZES selection of incomplete boxes as exemplars. For “nail polishes,” SAM-based point prompt segmentation in CountZES captures only partial objects; for “comic books,” point prompts isolate characters or cover details instead …

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

5 extracted references · 2 linked inside Pith

[4]

Countingdino: A training-free pipeline for class-agnostic counting using unsupervised backbones

Pacini, G., Bianchi, L., Ciampi, L., Messina, N., Amato, G., and Falchi, F. Countingdino: A training-free pipeline for class-agnostic counting using unsupervised backbones. arXiv preprint arXiv:2504.16570,

arXiv
[5]

Wimmer, T., Truong, P., Rakotosaona, M.-J., Oech- sle, M., Tombari, F., Schiele, B., and Lenssen, J. E. Anyup: Universal feature upsampling.arXiv preprint arXiv:2510.12764,

arXiv
[2018]

Can sam count anything? an empirical study on sam counting.arXiv preprint arXiv:2304.10817,

Ma, Z., Hong, X., and Shangguan, Q. Can sam count anything? an empirical study on sam counting.arXiv preprint arXiv:2304.10817,

arXiv
[2023]

Crafting papers on machine learning

Langley, P. Crafting papers on machine learning. In Langley, P. (ed.),Proceedings of the 17th International Conference on Machine Learning (ICML 2000), pp. 1207–1216, Stan- ford, CA,

2000
[2025]

Countr: Transformer-based generalised visual counting.arXiv preprint arXiv:2208.13721,

Liu, C., Zhong, Y ., Zisserman, A., and Xie, W. Countr: Transformer-based generalised visual counting.arXiv preprint arXiv:2208.13721,

arXiv

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.

[1] [4]

Countingdino: A training-free pipeline for class-agnostic counting using unsupervised backbones

Pacini, G., Bianchi, L., Ciampi, L., Messina, N., Amato, G., and Falchi, F. Countingdino: A training-free pipeline for class-agnostic counting using unsupervised backbones. arXiv preprint arXiv:2504.16570,

arXiv

[2] [5]

Wimmer, T., Truong, P., Rakotosaona, M.-J., Oech- sle, M., Tombari, F., Schiele, B., and Lenssen, J. E. Anyup: Universal feature upsampling.arXiv preprint arXiv:2510.12764,

arXiv

[3] [2018]

Can sam count anything? an empirical study on sam counting.arXiv preprint arXiv:2304.10817,

Ma, Z., Hong, X., and Shangguan, Q. Can sam count anything? an empirical study on sam counting.arXiv preprint arXiv:2304.10817,

arXiv

[4] [2023]

Crafting papers on machine learning

Langley, P. Crafting papers on machine learning. In Langley, P. (ed.),Proceedings of the 17th International Conference on Machine Learning (ICML 2000), pp. 1207–1216, Stan- ford, CA,

2000

[5] [2025]

Countr: Transformer-based generalised visual counting.arXiv preprint arXiv:2208.13721,

Liu, C., Zhong, Y ., Zisserman, A., and Xie, W. Countr: Transformer-based generalised visual counting.arXiv preprint arXiv:2208.13721,

arXiv