Pith. sign in

REVIEW 4 major objections 6 minor 23 references

Gloss-Free Representation Learning for Cross-Dataset Sign Spotting

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read Weak transcript-derived pseudo-glosses can pretrain a reusable Turkish Sign Language encoder that transfers to cross-dataset sign spotting, nearly doubling top-5 temporal localization IoU over raw spatial features.

desk verdict Weak supervision can help sign spotting in Turkish, but the headline gain is confounded by a missing architecture control; the shared-subset comparison is the real result. read the letter →

arxiv 2608.11332 v1 pith:ZHCYL2W4 submitted 2026-08-11 cs.CL cs.CV

classification cs.CLcs.CV
keywords TurkishSignLanguagespottingpseudo-glossesweaksupervisionrepresentationlearningtemporallocalizationbroadcastnewsagglutinativemorphology
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 asks whether noisy sentence-level transcripts, not manual glosses, can pretrain a reusable sign-language video encoder for Turkish Sign Language, a morphologically rich and resource-limited setting. It claims the answer is yes: an encoder shaped by weak pseudo-gloss labels derived from broadcast news transfers across datasets to localize isolated dictionary signs inside continuous sentence videos. The central comparison pits rule-based morphological lemmatization against a constrained LLM lexical normalizer, with the LLM version raising top-5 mean IoU from 0.235 to 0.465 and lifting the share of examples reaching an IoU of at least 0.50 from 23.9% to 56.2%. Because only the representation changes, not the spotting architecture, the result argues that text-derived pseudo-glosses carry genuine lexical-temporal signal.

What carries the argument

The load-bearing mechanism is pseudo-gloss pretraining with a two-temperature temporal-class scoring rule combined with normalized cross-correlation (NCC) matching. The sentence-level score for each pseudo-gloss aggregates per-frame similarities by taking a softmax over both pseudo-gloss classes and over time steps (Eq. 2); because no temporal target is supplied, any temporal ordering in the hidden states must emerge from the video dynamics, the encoder's inductive bias, and the weak label pressure. The second load-bearing component is the constrained LLM lexical-rule mapping, which selects candidate pseudo-glosses from a fixed inventory, canonicalizes inflectional and spelling variants while preserving derivational distinctions, and thereby raises overlap with the dictionary vocabulary from 52.89% to 69.76%.

What would settle it

Run the same word-to-pseudo-gloss mapping repeatedly on a held-out set of Turkish broadcast transcripts and measure the agreement rate; or evaluate the pretrained encoder on TSLD signs whose Turkish lexical forms never appear in TSL-News transcripts—if top-5 IoU collapses for those unseen signs, the representation is overfit to frequent transcript words rather than generalizing via sign structure.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that even noisy, sentence-level pseudo-glosses—bags of Turkish words extracted from transcripts without temporal boundaries—can shape a frozen-backbone video encoder into a lexical-temporal representation strong enough to support cross-dataset sign spotting. The encoder is pretrained on the new TSL-News broadcast corpus with a binary cross-entropy loss over pseudo-gloss presence, using a prototype head initialized from fastText subword embeddings, and then evaluated by normalized cross-correlation between isolated dictionary sign templates and continuous sentence hidden states on the new TSL-SB benchmark. The constrained LLM-assisted normalization, which maps surface forms onto a fixed vocabulary of 6,539 pseudo-glosses, outperforms both raw spatial features and the rule-based morphological lemmatization vocabulary of 4,802 on the full benchmark and on the 1,137 examples shared by both vocabularies. The frequency analysis (Spearman rho = -0.069) supports the claim that the gains are not mainly memorization of frequent pseudo-gloss labels.

Load-bearing premise

The frozen, manually reviewed LLM word-to-pseudo-gloss mapping is correct and consistent; the paper states that no independent repeat-call consistency audit of the original local LLM server was logged, so systematic misassignment of Turkish surface words could corrupt the training targets and make the reported transfer gains reflect text-matching artifacts rather than sign structure.

Editorial extensions

If this is right

  • Cross-dataset sign spotting becomes feasible for languages that have broadcast data but no expert gloss annotations: a single pretrained encoder can serve dictionary-query spotting on a held-out corpus.
  • Constrained LLM normalization over a fixed vocabulary is a viable alternative to morphological analyzers for agglutinative languages, buying both vocabulary coverage and localization accuracy.
  • Downstream translation starts from a better visual representation: reusing the pretrained encoder improves BLEU-4 from 9.60 to 11.04 and ROUGE from 23.48 to 27.43.
  • The weak supervision signal is chiefly lexical-temporal rather than frequency-based, since localization quality is nearly flat across train-sentence frequency quartiles.
  • The representation rather than the classifier carries the spotting signal: all-vocabulary local NCC outperforms direct pseudo-gloss score localization (top-5 IoU 0.516 versus 0.280).

Reading between the lines

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

  • A natural test that the paper does not run: feed the same encoder to a proper temporal-proposal generator and check whether the representation lifts open-vocabulary spotting, not just interval-constrained matching.
  • The closed-vocabulary LLM normalization could be replaced by an unsupervised subword segmenter, removing the need for a fixed candidate inventory and easing transfer to new broadcast domains without manual review.
  • The rule-based analyzer and the LLM mapper make different errors, so a hybrid that consults the analyzer only when the LLM rejects all candidates might reduce supervision noise further.
  • If the representation is truly lexical-temporal, it should be usable for sign-language retrieval and dictionary mining beyond spotting; the paper's future-work list points there but does not test retrieval.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper studies whether weak transcript-based supervision from broadcast news can pretrain a reusable sign-language video encoder under Turkish's agglutinative morphology. The authors introduce TSL-News, a Turkish broadcast corpus with sentence-level transcripts, and TSL-SB, a dictionary-based spotting benchmark built from the TSLD corpus. They compare two pseudo-gloss construction strategies from text: rule-based morphological lemmatization and a constrained LLM-assisted lexical normalization over a fixed vocabulary. The stage-1 encoder uses DINOv2 frame features with LoRA adaptation, a MetaFormer temporal encoder, and a fastText-initialized prototype head trained with a sentence-level BCE loss. Evaluation uses normalized cross-correlation (NCC) between isolated dictionary templates and continuous example video representations, reporting top-k temporal IoU. The LLM lexical-rule encoder is reported to improve top-5 mean IoU from 0.235 (raw spatial features) to 0.465, with 56.2% of examples reaching IoU > 0.50. A shared-subset comparison between the two pseudo-gloss strategies shows gains of about 0.10 IoU for the LLM strategy. The paper also reports a downstream translation check where pretraining improves BLEU-4 from 9.60 to 11.04. The central claim is that noisy sentence-level pseudo-glosses can shape a visual encoder into a lexical-temporal representation that transfers cross-dataset for sign spotting.

Significance. If the claims hold, the paper offers a practical weak-supervision route for low-resource sign languages and introduces two potentially useful resources (TSL-News and TSL-SB). The diagnostic protocol is thoughtful: the shared-subset comparison in Sec. 5.1 controls for vocabulary-coverage differences, the frequency analysis in Sec. 5.1 addresses memorization, and the multi-angle evaluation in Sec. 5.3 attempts to separate representation quality from classifier bias. The main unresolved threats are (1) the no-pretraining baseline lacks the temporal encoder and other architectural components, so the headline gain is not attributable to pseudo-gloss supervision; (2) the absence of a random-label or untrained same-architecture control; (3) single-seed reporting without variance; and (4) the benchmark only contains signs present in pretraining, limiting the cross-dataset claim. These are fixable with additional experiments, and the shared-subset comparison already provides partial evidence that label quality matters.

major comments (4)
  1. [Sec. 4.3, Table 3] The spatial-feature baseline used in Table 3 is not a controlled ablation of pseudo-gloss supervision: it lacks the trainable MetaFormer temporal encoder, temporal downsampling, and LoRA adaptation present in every pretrained encoder (Sec. 4.3, Appendix A.1). The reported gain from 0.235 to 0.465 top-5 IoU therefore conflates supervision with architectural and temporal-processing changes. To attribute the gain to pseudo-glosses, provide a same-architecture no-pretraining baseline, e.g., the MetaFormer encoder with random or frozen DINOv2 features evaluated without stage-1 training, or a control trained with shuffled/random pseudo-gloss labels. The shared-subset comparison in Sec. 5.1 controls coverage but not this architecture confound.
  2. [Sec. 5.3, Table 6] The pseudo-gloss classifier diagnostic achieves only 0.280 top-5 IoU versus 0.465 for target-known NCC on the same LLM lexical-rule benchmark. This large gap suggests that the dominant matching signal in NCC may be temporally smoothed visual similarity rather than lexical alignment. Please discuss this discrepancy explicitly and, if possible, report target-known NCC for the same-architecture no-training baseline. Such a comparison would help identify whether the NCC gain arises from learned lexical-temporal structure or from temporal smoothing alone.
  3. [Appendix A.1, Tables A.1 and A.2] All experiments use a single seed (seed 1). No variance or statistical significance measures are reported across Tables 3-7. Because the central claims are comparative improvements (LLM vs. morphology vs. no pretraining), single-seed results are insufficient to establish reliability. Please run at least three seeds with the same protocol and report means and standard deviations, or provide a clear justification for why the reported differences are beyond stochastic variation.
  4. [Sec. 5.1, Fig. 3] The benchmark by construction evaluates only signs that appear at least once in the TSL-News pretraining text, as the paper acknowledges in the Fig. 3 caveats. This limits the 'cross-dataset' claim to within-vocabulary transfer and leaves open what happens for unseen signs. Please state this limitation prominently in the abstract and conclusion, and consider adding a small held-out vocabulary test (e.g., signs present in TSL-SB but excluded from TSL-News text) to probe generalization to unseen lexical items.
minor comments (6)
  1. [Abstract, Sec. 1] The statement 'raises top-5 mean IoU from 0.235 to 0.465' should clarify that the 0.235 baseline uses raw spatial features without the stage-1 architecture; as written, the abstract implies a controlled comparison.
  2. [Appendix A.3] The admission that 'no independent repeat-call consistency audit of the original local LLM server was logged' is important; please add a sentence in the main text or supplement acknowledging this as a limitation of the LLM-based normalization, since the frozen mapping may still contain inconsistencies.
  3. [Fig. 3] The Spearman correlation of -0.069 is reported without a p-value or confidence interval; please report the p-value or use a permutation test to support the claim that frequency is not associated with localization quality.
  4. [Table 3] The row for the spatial-feature baseline uses the same example set as the LLM lexical-rule encoder (N=1817, Skip=25), but it does not share the architecture; a note in the table caption would prevent misinterpretation.
  5. [Sec. 3.1, Eq. (2)] The definition of p_g in Eq. (2) is clear in principle, but it would help to explicitly state that softmax_g is over the vocabulary dimension and softmax_t' over the time dimension, and that both use the same input s_{t',g} scaled by their respective temperatures.
  6. [Tables 3 and 6] The symbol N denotes example groups in Table 3 but segment occurrences in the all-vocabulary row of Table 6; please make the definition explicit in each table caption.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the pseudo-gloss pretraining result is not equivalent to its inputs by construction, though the headline baseline conflates architecture with supervision.

full rationale

The paper's central derivation is that sentence-level pseudo-gloss labels derived from Turkish transcripts can pretrain a visual encoder whose hidden states support cross-dataset sign spotting. The evaluation measures temporal IoU by sliding an isolated dictionary sign's hidden-state representation over a continuous example video using NCC and comparing peaks with human temporal annotations (Sec. 4.3). This outcome is not defined in terms of the pseudo-gloss labels: the labels provide only sentence-level text supervision, while localization requires the encoder to produce temporally and visually discriminative representations. The controlled shared-subset comparison between the morphology-lemma and LLM encoders (0.368 vs. 0.473 top-5 IoU) holds the architecture fixed and shows label quality affects localization, giving independent content to the claim. The Table 3 comparison against raw spatial features is indeed confounded by the temporal encoder, downsampling, and LoRA adaptation present only in the trained models, but this is an experimental-control weakness, not circularity: no quantity is fitted to the evaluation outcome or defined as the prediction. The paper itself acknowledges the benchmark is built from Turkish word-form overlap rather than expert sign glosses, that truly unseen signs are not tested, and that no independent repeat-call LLM consistency audit was logged (Sec. 6 and Appendix A.3). These limitations bound generalization but do not make the derivation circular. No load-bearing self-citation or uniqueness-imported-from-authors pattern is present. Overall circularity score: 0.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The pretraining and evaluation pipeline depends on several domain assumptions about transcript-sign alignment and about the validity of text-derived pseudo-glosses; the main hand-chosen settings are the trimming margins, loss weight, and LLM mapping constraints. No new conceptual entities such as particles or forces are introduced.

free parameters (4)
  • Class and time temperatures tau_c and tau_t = initial 0.1, dynamic during training
    Trainable sharpness parameters in Eq. (2); they shape the pseudo-gloss score distribution and are updated during pretraining.
  • Pseudo-gloss loss weight = 10.0
    Hand-set weight on the BCE loss in Eq. (3); changes the balance between positive and negative supervision and can affect how strongly temporal structure is learned.
  • Evaluation trimming margins = 0.5 s per side for sentence search region, 0.2 s per side for isolated signs
    Hand-chosen to remove non-signing idle frames; directly enters Eq. (4) and all reported IoU values, with no ablation provided.
  • LLM lexical-rule mapping settings = temperature 0.0, candidate limit 15, batch size 10, frozen 23,162-word mapping
    These constraints define the 6,539-class pseudo-gloss vocabulary; the mapping is frozen and manually reviewed, so its construction choices affect all downstream results.
assumptions (5)
  • domain assumption Transcript text and signing are loosely aligned at sentence level.
    Section 3.1 and Table 1; the pipeline receives only sentence-level pseudo-gloss bags, so the pretraining objective presupposes that the transcript describes what is signed in the paired video.
  • domain assumption Pseudo-gloss labels from text normalization correspond to visually realized signed units.
    Section 3.2 and Eq. (3); if text-derived labels do not match signs in the video, the encoder learns text artifacts rather than sign structure.
  • domain assumption DINOv2 frame features plus MetaFormer temporal encoding carry enough sign identity from RGB frames.
    Section 3.1 and Table A.1; no hand or face keypoints are used, so representation quality is bounded by these pretrained visual components.
  • domain assumption Fixed-length normalized cross-correlation sliding with hand-set trimming is a valid temporal localizer.
    Section 4.3, Eq. (4); signs in continuous signing vary in duration and co-articulation, so the fixed template length can bias IoU estimates.
  • domain assumption The 2004-entry TSLD pool and 596 annotated signs are representative enough to measure transfer.
    Section 4.2 and Table 2; TSL-News contains only three signers, and dictionary productions differ from broadcast signing, so the benchmark may not generalize to the full language.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Gloss-Free Representation Learning for Cross-Dataset Sign Spotting." pith.science (2026). https://pith.science/paper/ZHCYL2W4

@misc{pith2026260811332,
  author       = {Pith},
  title        = {Pith review of: Gloss-Free Representation Learning for Cross-Dataset Sign Spotting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZHCYL2W4}},
  note         = {Machine review of arXiv:2608.11332}
}
read the original abstract

Sign-language research for resource-constrained languages is often limited by the cost of dense linguistic labels such as glosses, temporal boundaries, and sign order. Broadcast news offers a practical alternative by pairing continuous signing with spoken-language transcripts, but this supervision is weak since text and signing are loosely aligned. Morphologically rich languages such as Turkish add further difficulty, as the same lexical meaning can appear in many inflected forms while some derived forms should remain distinct. We study whether weak transcript-based supervision can pretrain a reusable sign encoder in this setting, where poor text normalization can fragment pseudo-gloss targets and weaken representation learning. Unlike prior pseudo-gloss pipelines designed mainly to improve translation, we test whether the pretrained encoder transfers as a reusable representation for cross-dataset sign spotting. We pretrain on TSL-News, a new Turkish broadcast corpus, using pseudo-gloss labels derived from transcripts rather than manual annotation, comparing rule-based morphological lemmatization with constrained LLM-assisted normalization over a fixed vocabulary. We evaluate the learned representations via cross-dataset sign spotting on a new TSL Spotting Benchmark built from the TSL Dictionary corpus. The LLM-assisted encoder raises top-5 temporal localization mean IoU from 0.235 to 0.465, with 56.2% of examples reaching an IoU of at least 0.50; a frequency analysis suggests this gain is not mainly driven by memorizing frequent pseudo-gloss labels. In a downstream translation check, the same pretraining improves BLEU-4 from 9.60 to 11.04 and ROUGE from 23.48 to 27.43. These results show that loosely aligned broadcast data can provide effective weak supervision for learning sign representations that capture both lexical content and temporal structure.

Figures

Figures reproduced from arXiv: 2608.11332 by the authors.

Figure 1
Figure 1. Overview of the pretraining and translation pipeline. Turkish text is converted into pseudo-glosses via lemmatization; video frames pass through a frozen DINOv2 encoder and a trainable sign encoder, whose output feeds both the pseudo-gloss pre￾training head and, for context, a frozen GPT/XGLM decoder (both LoRA-adapted). Morphology-lemma strategy. The first strategy uses rule-based Turkish mor￾phological analysis, f… view at source ↗
Figure 2
Figure 2. Top-5 IoU survival curves for NCC localization. Each curve uses its own evalu￾ated set: spatial features (N=1817), morphology-lemma (N=1159), and LLM lexical￾rule (N=1817). The dashed line marks t=0.50, corresponding to the Top-5@0.50 col￾umn in [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. relates localization quality to the number of TSL-News train sen￾tences in which each target pseudo-gloss appears. The association is weak (\rho {=}-0.069 ), and the quartile means remain close even for low-frequency signs, sug￾gesting that the LLM lexical-rule encoder is not simply memorizing frequent pseudo-gloss labels. Two caveats apply. By construction, the benchmark con￾tains only signs that appear at least on… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Qualitative frame comparison for local all-vocabulary NCC. The top row shows the ground-truth Karışık (“complicated”) span, and the remaining rows show the top-5 isolated retrieval candidates. Similar hand and body configurations support that local NCC retrieves visual…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 17 canonical work pages

  1. [1]

    Transactions of the association for computational linguistics 5, 135–146 (2017) Gloss-Free Representation Learning for Sign Spotting 15

    Bojanowski, P., Grave, E., Joulin, A., Mikolov, T.: Enriching word vectors with subword information. Transactions of the association for computational linguistics 5, 135–146 (2017) Gloss-Free Representation Learning for Sign Spotting 15

  2. [2]

    Bragg, D., Koller, O., Bellard, M., Berke, L., Boudreault, P., Braffort, A., Caselli, N., Huenerfauth, M., Kacorri, H., Verhoef, T., et al.: Sign language recognition, generation,andtranslation:Aninterdisciplinaryperspective.In:Proceedingsofthe 21st international ACM SIGACCESS conference on computers and accessibility. pp. 16–31 (2019)

  3. [3]

    Universal Access in the Information Society23(3), 1305–1331 (2024)

    De Coster, M., Shterionov, D., Van Herreweghe, M., Dambre, J.: Machine trans- lation from signed to spoken languages: State of the art and challenges. Universal Access in the Information Society23(3), 1305–1331 (2024)

  4. [4]

    In: Adjunct Proceedings of the 25th ACM International Conference on Intelligent Virtual Agents, IVA Adjunct

    He, L.J., Sincan, O.M., Bowden, R.: Sign spotting disambiguation using large lan- guage models. In: Adjunct Proceedings of the 25th ACM International Conference on Intelligent Virtual Agents, IVA Adjunct. vol. 25 (2025)

  5. [5]

    In: 2025 IEEE 19th International Conference on Automatic Face and Gesture Recognition (FG)

    He, L.J., Walsh, H., Sincan, O.M., Bowden, R.: Hands-on: Segmenting individual signs from continuous sequences. In: 2025 IEEE 19th International Conference on Automatic Face and Gesture Recognition (FG). pp. 1–5. IEEE (2025)

  6. [6]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Hu, L., Gao, L., Liu, Z., Feng, W.: Continuous sign language recognition with correlation network. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 2529–2539 (2023)

  7. [7]

    arXiv preprint arXiv:2008.09918 (2020)

    Koller, O.: Quantitative survey of the state of the art in sign language recognition. arXiv preprint arXiv:2008.09918 (2020)

  8. [8]

    In: Vision interface

    Lewis, J.P.: Fast normalized cross-correlation. In: Vision interface. vol. 10, pp. 120–123 (1995)

Show all 23 references
  1. [9]

    In: Text sum- marization branches out

    Lin, C.Y.: Rouge: A package for automatic evaluation of summaries. In: Text sum- marization branches out. pp. 74–81 (2004)

  2. [10]

    (eds.): Güncel Türk İşaret Dili Sözlüğü

    Makaroğlu, B., Dikyuva, H. (eds.): Güncel Türk İşaret Dili Sözlüğü. Aile ve Sosyal Hizmetler Bakanlığı, Ankara (2017)

  3. [11]

    In: Proceedings of the Asian conference on computer vision (2020)

    Momeni, L., Varol, G., Albanie, S., Afouras, T., Zisserman, A.: Watch, read and lookup: learning to spot signs from multiple supervisors. In: Proceedings of the Asian conference on computer vision (2020)

  4. [12]

    In: Fourth Conference on Applied Natural Language Processing

    Oflazer, K., Guzey, C.: Spelling correction in agglutinative languages. In: Fourth Conference on Applied Natural Language Processing. pp. 194–195 (1994)

  5. [13]

    In: Fourth Conference on Applied Natural Language Processing

    Oflazer, K., Kuruoz, I.: Tagging and morphological disambiguation of turkish text. In: Fourth Conference on Applied Natural Language Processing. pp. 144– 149 (1994)

  6. [14]

    arXiv preprint arXiv:2304.07193 (2023)

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

  7. [15]

    In: Proceedings of the 40th annual meeting of the Association for Computational Linguistics

    Papineni, K., Roukos, S., Ward, T., Zhu, W.J.: Bleu: a method for automatic evaluation of machine translation. In: Proceedings of the 40th annual meeting of the Association for Computational Linguistics. pp. 311–318 (2002)

  8. [16]

    arXiv preprint arXiv:2405.10266 (2024)

    Raude, C., Prajwal, K., Momeni, L., Bull, H., Albanie, S., Zisserman, A., Varol, G.: A tale of two languages: Large-vocabulary continuous sign language recognition from spoken language supervision. arXiv preprint arXiv:2405.10266 (2024)

  9. [17]

    Artificial Intelligence Review57(10), 271 (2024)

    Shahin, N., Ismail, L.: From rule-based models to deep learning transformers archi- tecturesfornaturallanguageprocessingandsignlanguagetranslationsystems:sur- vey, taxonomy and performance evaluation. Artificial Intelligence Review57(10), 271 (2024)

  10. [18]

    In: 2023 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW)

    Walsh, H., Sincan, O.M., Saunders, B., Bowden, R.: Gloss alignment using word embeddings. In: 2023 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW). pp. 1–5. IEEE (2023) 16 O. A. Tüfekcioğlu et al

  11. [19]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Wong, R., Camgoz, N.C., Bowden, R.: Learnt contrastive concept embeddings for sign recognition. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 1945–1954 (2023)

  12. [20]

    arXiv preprint arXiv:2405.04164 (2024)

    Wong, R., Camgoz, N.C., Bowden, R.: Sign2gpt: Leveraging large language models for gloss-free sign language translation. arXiv preprint arXiv:2405.04164 (2024)

  13. [21]

    In: Proceedings of the IEEE/CVF international conference on com- puter vision

    Wong, R., Camgoz, N.C., Bowden, R.: Signrep: Enhancing self-supervised sign rep- resentations. In: Proceedings of the IEEE/CVF international conference on com- puter vision. pp. 22804–22814 (2025)

  14. [22]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Yu,W.,Luo,M.,Zhou,P.,Si,C.,Zhou,Y.,Wang,X.,Feng,J.,Yan,S.:Metaformer is actually what you need for vision. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 10819–10829 (2022)

  15. [23]

    complicated

    Zhou, B., Chen, Z., Clapés, A., Wan, J., Liang, Y., Escalera, S., Lei, Z., Zhang, D.: Gloss-free sign language translation: Improving from visual-language pretraining. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 20871–20881 (2023) Gloss-Fre...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.