Pith. sign in

REVIEW 4 major objections 3 minor 25 references

Multi-Scale ViT Inference with Habitat-Fit Priors and kNN Retrieval for Multi-Species Plant Identification

T0 review · 4 major / 3 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read On the PlantCLEF 2026 multi-species plant identification benchmark, the paper claims that the largest performance gains come not from extra training data or more complex training schemes, but from inference-time structure: multi-scale tile

desk verdict Solid, honest systems paper, but the appendix undercuts the main causal claim: habitat-fit centroids come from test-set predictions, making the 'training-data prior' partly transductive. read the letter →

arxiv 2607.14509 v1 pith:VJXO6LAW submitted 2026-07-16 cs.CV cs.AIcs.LG

Alper Erten , Murilo Gustineli , Adrian Cheung This is my paper · ORCID
classification cs.CVcs.AIcs.LG
keywords PlantCLEF2026multi-labelidentificationDINOv2multi-scaletileinferencehabitat-fitdemotionkNNretrievaltemporalfusionecologicalpriors
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

This working note claims that on the PlantCLEF 2026 multi-species plant-identification benchmark, the winning moves were not more training data or fancier training losses but inference-time structure: tile the quadrat at several scales, pool the strongest per-species evidence, blend in kNN retrieval, and then reweight predictions by where the species actually grows. The paper's ablations attribute the largest single gain to a habitat-fit demotion that multiplies classifier probabilities by a per-source geographic-and-altitude fit factor. If that attribution holds, dataset-structure priors are a more cost-effective lever than model or training complexity on severe single-to-multi-label domain shift. The selected submission ranked third (private macro-F1 0.43902), while an unselected configuration of the same pipeline scored 0.45777.

What carries the argument

The habitat-fit demotion factor f_{s,c}=geo_{s,c}·alt_{s,c}: a per-source geographic score counting training observations of species c within 250 km of the source centroid, times an altitude score penalizing species whose median training elevation falls outside the source band. It is what carries the largest ablation delta. Around it: multi-scale max-pooled tile aggregation (86 tiles, T=1.5), a FAISS kNN blend (beta=0.70 multi-visit), source-aware temporal fusion (max or similarity-weighted, alpha=0.15/0.30), and a two-step admission rule (threshold 0.085, top-10, floor 3 with ratio 0.5).

What would settle it

Re-run the cumulative ablation with 5-10 random seeds or re-tune beta_multi, alpha_RNNB, n_floor, r and tau on independent 11% splits; if habitat-fit removal's private delta shrinks below about -0.02 or changes sign, the central attribution claim collapses.

Watch

Extended reading notes

Core claim

The paper establishes that a ViT-L/14 fine-tuned on 1.4M single-label plant images can be turned into a strong multi-label quadrat predictor by a fixed inference recipe: centre-crop each ~3000px quadrat, tile at 3x3/4x4/5x5/6x6 (86 tiles), max-pool tile softmaxes at temperature 1.5, blend with a FAISS kNN distribution over 860k ArcFace embeddings, fuse predictions across repeated visits per collection source, then apply habitat-fit demotion (geographic proximity within 250 km and altitude-band compatibility per source) and a South-Western Europe mask. On the private leaderboard the full pipeline scores 0.45777; removing habitat-fit costs -0.04075, removing kNN costs -0.01256, collapsing to o

Load-bearing premise

The causal attribution to habitat-fit demotion rests on a single ablation run whose downstream scalars were tuned on the 11% public subset, so the -0.04075 private delta may partly be selection noise rather than true component value.

Editorial extensions

If this is right

  • Habitat-fit and ecological masking are worth more than the kNN blend or any single extra scale, so future PlantCLEF systems should spend effort on train-set metadata priors before adding models.
  • The selected scale set (3+4+5+6) is stable across backbones, meaning the optimal tiling is set by image structure, not model capacity.
  • Public-leaderboard tuning is unsafe: larger scale sets look better publicly but degrade privately, and the best private run was not selected because selection used public scores.
  • Training-centric adaptation using LUCAS or SAM crops can actively hurt because of distribution mismatch; inference-time post-processing on cached in-domain features is the safer lever.
  • The null TileQ-Decoder result shows label-as-query decoding from synthetic CLS pseudo-quadrats fails under sparse FAISS pseudo-labels.

Reading between the lines

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

  • A testable extension is to replace source-level centroids with per-quadrat coordinates (or predicted locations), which would let habitat-fit priors act at the individual-plot level rather than the coarser source level.
  • Because all scalar post-processing knobs were tuned on the 11% public split, the true component contributions are likely smaller than reported; a randomized hyperparameter search would bound the selection bias.
  • The unimodal similarity distribution that killed Sampler B suggests FAISS pseudo-labelling needs a calibrated threshold or a density-aware relabelling scheme, transferable to other sparse-retrieval supervision settings.
  • If the attribution replicates, the same inference recipe should transfer directly to other geo-tagged multi-label remote-sensing tasks where single-label training data exist.
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.

Referee Report

4 major / 3 minor

Summary. This paper describes the DS@GT ARC entry to PlantCLEF 2026, a third-place Kaggle solution for multi-species plant identification in high-resolution quadrat images. The pipeline uses a fine-tuned DINOv2 ViT-L/14 on a multi-scale tiling (3×3 to 6×6), blends tile-level softmax with a FAISS kNN retrieval distribution, performs per-source temporal fusion for multi-visit quadrats, applies a habitat-fit demotion (geographic + altitude scores), a South-Western Europe mask, and a two-stage admission rule. The authors report a private macro-F1 of 0.43902 for the selected submission and 0.45777 for an unselected config used as baseline; cumulative ablations attribute the largest gain (−0.04075) to removing habitat-fit/masks, followed by kNN and multi-scale aggregation. Three training-centric or inference-augmentation directions (cross-region transformer + noisy-student distillation, TileQ-Decoder, SAM instance crops) gave null results. Code and configuration files are released.

Significance. If the attribution holds, the paper is a useful data point: on this benchmark, structural priors about the test collection's source geography and multi-scale aggregation beat extra training data or learned decoders. The paper is unusually transparent: it reports null results with pre-registered hypotheses, documents public-leaderboard tuning, and releases code. However, the central attribution is currently not cleanly established because (i) the largest ablation bundles several mechanisms and (ii) the habitat-fit centroids appear to be derived from an earlier model's predictions on the test set, which conflicts with the paper's claim that those priors come entirely from training data. These issues are addressable with additional disclosure and ablations.

major comments (4)
  1. [§4.5 / Appendix A Step 2 / §6] The paper twice asserts that habitat-fit priors are 'derived entirely from training-data statistics' (§6) and 'not hardcoded inputs' (Appendix A). However, Appendix A Step 2 defines the source centroids using 'top-30 most frequently predicted species from an earlier locked pipeline stage' on the test sources. Since Eq. (11) uses distances to these centroids, the habitat-fit factor implicitly carries test-set prediction statistics. This is transductive use of the test distribution, not a purely training-derived prior. The −0.04075 private-F1 delta in Table 2 therefore conflates a genuine ecological prior with a test-set-composition signal. Please either (a) recompute centroids from training metadata alone (e.g., median coordinates of training observations per source) and re-run Abl-2, or (b) clearly re-frame the claim and disclose the test-set dependence.
  2. [§5.2 / Table 2 / §6] Abl-2 is labeled 'habitat-fit demotion' but §6 says the ablation removes 'the geographic score, altitude score, and ecological masks together.' It therefore also removes the South-Western Europe mask and likely changes the admission step (since masks alter the candidate pool). Thus the largest observed delta (−0.04075) is not the isolated contribution of Eqs. (9)–(12). To support the paper's main claim, report separate ablations: habitat-fit only (with mask fixed), mask only, and admission/count-bound only.
  3. [§5.2 / Table 2] All ablation differences are single-run private-leaderboard measurements with no variance estimates. The paper itself documents substantial public/private divergence (Table 3: scale 7 has public 0.40523 but private 0.38826), so a −0.04075 difference from one configuration could be partly selection noise. Please state the number of runs per configuration and, if feasible, provide repeated predictions (e.g., by varying the random seed in the pipeline's stochastic stages) or at least a sensitivity analysis over the choice of baseline (the unselected 0.45777 vs selected 0.43902).
  4. [§6] The claim that the overfitting risk 'is specific to the admission and blending hyperparameters' and not to habitat-fit is not supported. The habitat-fit step has tunable scalars: β=0.10, n_near_min=40, the ±50% altitude band tolerance, the 1000 m distance scale, and the country set used for the geographic mask. The paper gives no evidence that these were fixed before observing the public leaderboard. If they were selected using the public subset, the same overfitting caveat applies to the largest component; if they were not, please say so explicitly.
minor comments (3)
  1. [§5.2 / Table 2] The row label 'habitat-fit demotion' should match the text 'habitat-fit + ecological masks'; clarify in the table caption what exactly is removed in each Abl row.
  2. [§4.3] The notation β in Eq. (4) is later specialized to β_multi and β_single; define the visit-aware schedule before Eq. (4) to avoid ambiguity.
  3. [Abstract / §6] The abstract states 'habitat-fit demotion and multi-scale aggregation are the largest individual contributors'; given the bundling in Abl-2, rephrase to 'the combination of habitat-fit and geographic masking'.

Circularity Check

1 steps flagged · score 6.0 of 10

Habitat-fit's central ablation gain is partly circular: its source centroids are fitted to the test set's own model predictions, though the paper calls the priors 'derived entirely from training-data statistics.'

  1. fitted input called prediction [Appendix A Step 2; §4.5 Eqs. 9–12; §6]
    "Initial hand-set centroids were replaced with data-derived centroids ... For each of the eight test sources, the script identifies anchor species as the top-30 most frequently predicted species from an earlier locked pipeline stage ... The refined centroid is then computed as the median latitude and longitude of those anchor species’ training observations ... We note that the habitat-fit priors and geographic mask are derived entirely from training-data statistics and are therefore not subject to this concern."

    The source centroids computed in Appendix A Step 2 are inputs to the geographic score geo_{s,c} in Eq. (11) and so to the habitat-fit factor f_{s,c}=geo*alt in Eq. (10), which rescales each quadrat's fused probabilities in Eq. (9). Because the anchor species are chosen as the top-30 species most frequently predicted on the test quadrats by an earlier locked pipeline stage, the habitat-fit factor is not a training-data-only prior: it is fitted to the very test images whose predictions it then modifies. The −0.04075 private-F1 delta attributed to habitat-fit in Table 2 / Abl-2 therefore conflates a genuine training-metadata prior with a transductive fit to the evaluation set. The §6 statement that habitat-fit priors are 'derived entirely from training-data statistics' is contradicted by the

full rationale

This is an empirical competition paper without a formal derivation chain, so most components are not circular: the ViT classifier, kNN retrieval, multi-scale tiling, temporal fusion, and the external-leaderboard ablations are tested against data, and the three null-result directions (TileQ-Decoder, LUCAS distillation, SAM crops) are independent. The material circularity is confined to habitat-fit, which the paper itself identifies as the largest contributor. Appendix A Step 2 shows the per-source centroids — which determine the geographic score of the habitat-fit factor — are derived from the top-30 species most frequently predicted on the test quadrats by an earlier pipeline version. That makes the habitat-fit prior partly a function of the evaluation set, directly contradicting the paper's claim in §6 that it is 'derived entirely from training-data statistics.' The central ablation result (−0.04075 private F1 when habitat-fit is removed) is therefore not a clean measurement of a training-derived ecological prior; it is partly attributable to a transductive fit to the test images. This is partial circularity in the central claim, not a full derivation-equivalence, because the prior also includes genuine training-metadata statistics. A separate, non-circular confound is that Abl-2 removes the geographic mask and admission rule together with habitat-fit, which weakens attribution but is not circularity.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The central claims rest on several fitted scalars and domain assumptions about metadata reliability, test-set ID structure, and single-run ablation validity. There are no invented physical entities; habitat-fit priors are post-processing statistics computed from public training metadata and test predictions.

free parameters (8)
  • kNN blend weights (β_multi, β_single) = 0.70, 1.00
    Blend classifier and retrieval at tile level (Eq. 4); tuned on public leaderboard; kNN disabled for single-visit images.
  • Temporal fusion coefficients (α_RNNB, α_default) = 0.30, 0.15
    Mixing per-quadrat prediction with location prior (Eq. 5); source-dependent and tuned.
  • Admission rule (n_floor, r, threshold, top_k) = n_floor=3, r=0.5, threshold=0.085, top_k=10
    Two-stage prediction admission (Sec. 4.6) tuned on public leaderboard.
  • kNN retrieval temperature τ = 0.07
    Softmax over neighbor similarities (Eq. 3); tuned.
  • Classifier softmax temperature T = 1.5
    Applied to per-tile logits before max-pooling (Eq. 2); hand-set.
  • Geographic score β and n_near_min = 0.10, 40
    Floor and saturation for geographic penalty (Eq. 11); hand-set.
  • Altitude score β, distance scale, band tolerance = β=0.10, 1/1000 per m, ±50% band
    Penalty schedule for altitude mismatch (Eq. 12); hand-set.
  • Multi-scale set S = {3,4,5,6}
    Chosen via sweep; private best, public divergent (Table 3).
assumptions (5)
  • domain assumption DINOv2 ViT features transfer to fine-grained plant identification and the 2024 fine-tuned checkpoint is a valid base
    Sec. 3.2 adopts organizer-provided ViT-B and fine-tunes ViT-L; all results depend on this transfer.
  • domain assumption Test quadrat_id strings reliably encode visit/location and source, and siblings are same physical plot
    Sec. 4.4 uses trailing date suffix removal and source prefix for temporal fusion and per-source priors.
  • domain assumption Training-set latitude/longitude/altitude metadata are accurate enough to compute centroids and altitude bands
    Sec. 4.5 and Appendix A use PlantCLEF 2024 metadata; species without coordinates get neutral factor 1.0.
  • ad hoc to paper Single-run leaderboard deltas approximate component contributions
    Table 2 ablations have no repeats/error bars; paper treats deltas as causal.
  • domain assumption Public leaderboard feedback is a valid selection signal for post-hoc scalars
    Development used ~11% public scores to set β, α, τ, floors; paper notes this may not generalize.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-Scale ViT Inference with Habitat-Fit Priors and kNN Retrieval for Multi-Species Plant Identification." pith.science (2026). https://pith.science/paper/VJXO6LAW

@misc{pith2026260714509,
  author       = {Pith},
  title        = {Pith review of: Multi-Scale ViT Inference with Habitat-Fit Priors and kNN Retrieval for Multi-Species Plant Identification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VJXO6LAW}},
  note         = {Machine review of arXiv:2607.14509}
}
read the original abstract

This paper describes DS@GT ARC's third-place solution to the PlantCLEF 2026 challenge on multi-species plant identification in vegetation quadrat images, where systems must predict every species present in high-resolution (~3000 x 3000 pixel) plot photographs while training only on single-label images of individual plants. The pipeline is built around a fine-tuned DINOv2 ViT-L/14 classifier applied over a multi-scale tile decomposition of each quadrat, with per-tile predictions blended with a FAISS kNN retriever and post-processed by source-aware temporal fusion across repeated plot visits, a habitat-fit demotion that injects geographic and altitude priors from the training data, and a South-Western Europe geographic mask. Habitat-fit demotion and multi-scale aggregation are the largest individual contributors in the ablations. Two complementary training-centric directions, a cross-region transformer with noisy-student distillation on the LUCAS dataset and a label-as-query transformer decoder over synthetic CLS-domain pseudo-quadrats, yielded null results. An inference-time augmentation with instance-aware segmentation crops also did not improve performance. The selected submission reaches a private-leaderboard macro-F1 of 0.43902 (third place; public 0.51096); an unselected configuration of the same pipeline scored above 0.45 on the private set. Code: https://github.com/dsgt-arc/plantclef-2026.

Figures

Figures reproduced from arXiv: 2607.14509 by the authors.

Figure 1
Figure 1. Twelve single-label training images displaying the following six species: Veronica urticifolia Jacq., Tripleurospermum inodorum (L.) Sch.Bip., Geum reptans L., Carex spicata Huds., Oenothera rosea L’Hér. ex Aiton, Lamium bifidum Cirillo. The challenge test set is identical to the 2025 set: 2,105 high-resolution (typically ≈3000×3000 pixel) top-down photographs of 0.5×0.5 m quadrats produced by botanical experts acro… view at source ↗
Figure 2
Figure 2. Subset of twelve test set images showcasing the significant domain shift between different quadrats. Evaluation metric. The official metric is a hierarchical macro-averaged sample 𝐹1 score with set￾based per-image comparison [3]. For each test image 𝑖, precision and recall are computed by treating the predicted and true species sets as binary multi-labels, and an image-level 𝐹1𝑖 is computed in the usual way. Image-l… view at source ↗
Figure 3
Figure 3. Multiscale tile aggregation pipeline. A quadrat image is centre-cropped, tiled at multiple scales, processed by a ViT-L/14 classifier, and aggregated via max-pooling to produce the final per-image prediction. The final pipeline uses the scale set 𝑆 = {3, 4, 5, 6}, resulting in 9 + 16 + 25 + 36 = 86 tiles per image. This configuration was selected based on the multi-scale sweep in Section 5 ( [PITH_FULL_IMAGE:figure… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: FAISS kNN retrieval pipeline: ViT-L + ArcFace embeddings are used to perform cosine similarity search over a FAISS IndexFlatIP index. The top-𝑘 nearest neighbors are retrieved, and their similarities are converted into weights to produce a class-level vote distribution…
Figure 5
Figure 5. Figure 5: Number of quadrats per source, separated by multi-visit and single-visit sites. Most samples belong to multi-visit collections, while several sources (e.g., LISAH, GUARDEN) contain predominantly single-visit data, motivating the use of source-aware temporal fusion. For…
Figure 6
Figure 6. Figure 6: Habitat-fit demotion. The fused per-quadrat probabilities are rescaled by a source-dependent fit factor 𝑓𝑠,𝑐 that encodes geographic and altitude priors. The geographic score reflects species presence within a 250 km radius of the source centroid, while the altitude sc…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 1 canonical work pages

  1. [1]

    Martellucci, I

    G. Martellucci, I. Moummad, H. Goëau, P. Bonnet, F. Vinatier, A. Joly, Overview of PlantCLEF 2026: Identify multi-species plants in images of vegetation plots, in: Working Notes of CLEF 2026 – Conference and Labs of the Evaluation Forum, 2026

    2026

  2. [2]

    Picek, L

    L. Picek, L. Adam, S. Kahl, R. Bossy, L. Chrobak, H. Goëau, K. Papafitsoros, H. Klinck, W.-P. Vellinga, R. Planqué, T. Denton, K. Barnard, C. Nédellec, L. Deléger, M. Courtin, G. Martellucci, I. Moummad, F. Vinatier, P. Bonnet, A. Joly, Overview of LifeCLEF 2026: Ai challenges for biodiversity understanding and ecosystem management, in: International Conf...

    2026

  3. [3]

    Martellucci, H

    G. Martellucci, H. Goëau, P. Bonnet, F. Vinatier, A. Joly, Overview of PlantCLEF 2025: Multi-species plant identification in vegetation quadrat images, in: CLEF (Working Notes), 2025

    2025

  4. [4]

    Goeau, V

    H. Goeau, V. Espitalier, P. Bonnet, A. Joly, Overview of PlantCLEF 2024: Multi-species plant identification in vegetation plot images, in: CLEF (Working Notes), 2024

    2024

  5. [5]

    d’Andrimont, M

    R. d’Andrimont, M. Yordanov, L. Martinez-Sanchez, P. Haub, O. Buck, C. Haub, B. Eiselt, M. van der Velde, Lucas cover photos 2006–2018 over the eu: 874 646 spatially distributed geo-tagged close-up photos with land cover and plant species label, Earth System Science Data 14 (2022) 4463–4472

    2006

  6. [6]

    Carion, L

    N. Carion, L. Gustafson, Y.-T. Hu, S. Debnath, R. Hu, D. Suris, C. Ryali, K. V. Alwala, H. Khedr, A. Huang, et al., Sam 3: Segment anything with concepts, arXiv preprint arXiv:2511.16719 (2025)

    arXiv 2025

  7. [7]

    Gustineli, A

    M. Gustineli, A. Miyaguchi, A. Cheung, D. Khattak, Tile-based vit inference with visual-cluster priors for zero-shot multi-species plant identification, arXiv preprint arXiv:2507.06093 (2025)

    arXiv 2025

  8. [8]

    A. Joly, H. Goëau, H. Glotin, C. Spampinato, P. Bonnet, W.-P. Vellinga, J.-C. Lombardo, R. Planqué, S. Palazzo, H. Müller, Biodiversity information retrieval through large scale content-based identifi- cation: a long-term evaluation, in: Information Retrieval Evaluation in a Changing World: lessons learned from 20 years of CLEF, Springer, 2019, pp. 389–413

    2019

Show all 25 references
  1. [9]

    Goëau, P

    H. Goëau, P. Bonnet, A. Joly, Overview of plantclef 2023: Image-based plant identification at global scale, in: CLEF (Working Notes), 2023, pp. 1972–1981

    2023

  2. [10]

    Goëau, P

    H. Goëau, P. Bonnet, A. Joly, Plant identification based on noisy web data: The amazing perfor- mance of deep learning (LifeCLEF 2017), in: CLEF (Working Notes), 2017

    2017

  3. [11]

    Goëau, P

    H. Goëau, P. Bonnet, A. Joly, Overview of PlantCLEF 2022: Image-based plant identification at global scale, in: CLEF (Working Notes), 2022

    2022

  4. [12]

    S. Foy, S. McLoughlin, Utilising dinov2 for domain adaptation in vegetation plot analysis., in: CLEF (Working Notes), 2024, pp. 2083–2092

    2024

  5. [13]

    Kirillov, E

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, W.-Y. Lo, et al., Segment anything, in: Proceedings of the IEEE/CVF international conference on computer vision, 2023, pp. 4015–4026

    2023

  6. [14]

    Chulif, H

    S. Chulif, H. A. Ishrat, Y. L. Chang, S. H. Lee, Patch-wise inference using pre-trained vision transformers: Neuon submission to plantclef 2024., in: CLEF (Working Notes), 2024, pp. 2027– 2043

    2024

  7. [15]

    Takahashi, T

    R. Takahashi, T. Matsubara, K. Uehara, Data augmentation using random image cropping and patching for deep cnns, IEEE Transactions on Circuits and Systems for Video Technology 30 (2019) 2917–2931

    2019

  8. [16]

    Gustineli, A

    M. Gustineli, A. Miyaguchi, I. Stalter, Multi-label plant species classification with self-supervised vision transformers, arXiv preprint arXiv:2407.06298 (2024)

    2024 arXiv

  9. [17]

    Espitalier, Preprocessing is all you need: Theheartofnoise submission to plantclef 2025, CLEF 2025 Working Notes (2025) 1–22

    V. Espitalier, Preprocessing is all you need: Theheartofnoise submission to plantclef 2025, CLEF 2025 Working Notes (2025) 1–22

    2025

  10. [18]

    Herasimchyk, R

    H. Herasimchyk, R. Labryga, T. Prusina, Multi-label plant species prediction with metadata- enhanced multi-head vision transformers, arXiv preprint arXiv:2508.10457 (2025)

    2025 arXiv

  11. [19]

    Dengler, F

    J. Dengler, F. Jansen, O. Chusova, E. Hüllbusch, M. P. Nobis, K. Van Meerbeek, I. Axmanová, H. H. Bruun, M. Chytr`y, R. Guarino, et al., Ecological indicator values for europe (eive) 1.0, Vegetation Classification and Survey 4 (2023) 7–29

    2023

  12. [20]

    Oquab, T

    M. Oquab, T. Darcet, T. Moutakanni, H. Vo, M. Szafraniec, V. Khalidov, P. Fernandez, D. Haziza, F. Massa, A. El-Nouby, et al., Dinov2: Learning robust visual features without supervision, arXiv preprint arXiv:2304.07193 (2023)

    2023 arXiv

  13. [21]

    Darcet, M

    T. Darcet, M. Oquab, J. Mairal, P. Bojanowski, Vision transformers need registers, in: International Conference on Learning Representations (ICLR), 2024

    2024

  14. [22]

    Goëau, J.-C

    H. Goëau, J.-C. Lombardo, A. Affouard, V. Espitalier, P. Bonnet, A. Joly, PlantCLEF 2024 pretrained models on the flora of south-western europe based on a subset of Pl@ntNet collaborative images and a ViT base patch 14 DINOv2, Zenodo. https://doi.org/10.5281/zenodo.10848263, 2024

    2024 doi

  15. [23]

    Ridnik, G

    T. Ridnik, G. Sharir, A. Ben-Cohen, E. Ben-Baruch, A. Noy, Ml-decoder: Scalable and versatile classification head, in: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2023, pp. 32–41

    2023

  16. [24]

    Ridnik, E

    T. Ridnik, E. Ben-Baruch, N. Zamir, A. Noy, I. Friedman, M. Protter, L. Zelnik-Manor, Asymmetric loss for multi-label classification, in: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2021, pp. 82–91

    2021

  17. [25]

    URL: http://www

    PACE, Partnership for an Advanced Computing Environment (PACE), 2017. URL: http://www. pace.gatech.edu. A. Habitat-Fit Source Prior Computation The habitat-fit factors used in Section 4 are not hardcoded constants. They are the output of a two-step computation performed once o...

    2017

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.