Pith. sign in

REVIEW 4 major objections 4 minor 32 references

HW-MLVQA: Elucidating Multilingual Handwritten Document Understanding with a Comprehensive VQA Benchmark

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read This paper presents HW-MLVQA, the first bilingual handwritten VQA benchmark with evidence grounding, and shows current vision-language models localize answers near chance.

desk verdict First bilingual handwritten VQA benchmark with grounding, but the paper needs a data release, consistent stats, and a box-reliability study before its results can be trusted. read the letter →

arxiv 2507.15655 v1 pith:RGWNUCS5 submitted 2025-07-21 cs.CV

Aniket Pal , Ajoy Mondal , Minesh Mathew , C.V. Jawahar This is my paper · ORCID
classification cs.CV
keywords HW-MLVQAhandwrittendocumentunderstandingmultilingualvisualquestionansweringevidencegroundingOCRHindi-Englishbenchmarkvision-languagemodels
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 paper introduces HW-MLVQA, a benchmark of 2,400 English and Hindi question-answer pairs grounded in 1,520 handwritten pages, where a correct answer must be accompanied by the bounding box of the evidence in the handwriting. The authors evaluate LLaMA 3.1 and Qwen2VL under text-only, image-only, and combined inputs, with OCR from Google and EasyOCR, and they report that a state-of-the-art vision-language model such as Qwen2VL extracts little linguistic structure from handwriting alone in zero-shot settings. The key reported result is that evidence localization is essentially unsolved: mean intersection-over-union is 0.0166 for English and 0.0126 for Hindi, near zero. If the benchmark is sound, a sympathetic reader would conclude that handwritten multilingual document understanding is an open problem and that OCR quality, especially for open-source tools, is a primary bottleneck.

What carries the argument

The load-bearing component is the Evidence-Based Grounded VQA (EB-GVQA) task together with the benchmark's construction pipeline. EB-GVQA formalizes outputs as an answer string plus a set of evidence pairs (bounding box, page index), scored with ANLS for the answer and mean IoU for the evidence. The dataset starts from 400 English and 400 Hindi contexts drawn from SQuAD and MLQA, splits each context into about 50-word segments, and has volunteers handwrite them onto A4 pages; an automatic annotation stage then uses Google OCR and EasyOCR to extract words, matches the OCR tokens against the expected answer words from the QA files, and writes XML boxes that a five-to-six person team manually verifies.

What would settle it

Take a random sample of HW-MLVQA pages, have independent human annotators draw the exact bounding boxes of the answer spans, and compute the IoU between the benchmark's XML boxes and the human boxes; if the agreement is well below near-perfect (for example, mean IoU below 0.9), then the near-zero VLM grounding numbers partly reflect noise in the labels rather than pure model failure.

Watch

Extended reading notes

Core claim

HW-MLVQA is claimed to be the first benchmark that combines multilingual (English and Hindi) handwritten documents with evidence-based grounded VQA: each question's ground truth includes both an answer string and bounding boxes marking where in the page the answer is written. The paper's central experimental claim is that current models cannot solve this task. Under zero-shot evaluation, the paper's Table 4 reports that Qwen2VL-7B with image-only input reaches an exact-match accuracy of 57.51% on English and 22.70% on Hindi, and its evidence localization is effectively absent, with mean IoU of 0.0166 on English and 0.0126 on Hindi and variance below 0.002. The authors conclude that state-of-the-art vision-language models struggle to capture linguistic structures from handwritten images alone, and that OCR noise, especially from open-source EasyOCR, further degrades performance when transcriptions are provided.

Load-bearing premise

The assumption that the answer-location boxes in the benchmark are accurate labels, even though they are produced by matching OCR output to answer words and EasyOCR's word accuracy on English pages is only 5.56%, so many answer boxes may be missing or misplaced.

Editorial extensions

If this is right

  • Any future system claiming to handle handwritten documents should be measured on both answer accuracy (ANLS or exact match) and evidence localization (IoU), with the benchmark's near-zero IoU numbers as the explicit target to beat.
  • Since Qwen2VL's exact match on English drops from 67% with ground-truth text to about 47% with GoogleOCR output, handwriting OCR quality is a direct bottleneck; improving OCR word accuracy should transfer to higher VQA scores.
  • The English-Hindi gap (image-only exact match 57.51% versus 22.70%) quantifies how much harder the second script is for a monolithic VLM, supporting the paper's call for language-specific adaptation.
  • Combined image-and-text inputs consistently outperform either modality alone (with ground truth, English exact match rises from 57.51% image-only to 68.00%), suggesting that future systems should fuse visual handwriting cues with noisy transcripts.

Reading between the lines

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

  • The paper does not compare its mean IoU values against a random-box baseline; a quantitative test of the 'near chance' reading would sample random boxes on the same pages and compute their IoU distribution.
  • Because the ground-truth boxes are produced by an OCR-matching pipeline in which EasyOCR's word accuracy is 5.56% on English pages, a subset of evidence labels is likely imperfect; a human-annotated sample would let users separate label noise from genuine model failure.
  • The abstract states 1,600 handwritten pages, while Section 3.4 and Table 1 give 1,520 (834 English plus 686 Hindi); the released dataset's actual page count should be confirmed before relying on scale claims.
  • In Section 5.2.3 the prose reports an image-only English exact match of 45.39% for Qwen2VL, while Table 4 reports 57.51% for the same setting; readers should verify which number matches the released evaluation.
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 / 4 minor

Summary. The paper presents HW-MLVQA, a benchmark for multilingual handwritten document VQA. Contexts are taken from SQuAD and MLQA, written by hand by volunteers, and paired with question-answer sets. The authors evaluate LLaMA 3.1 and Qwen2VL in text-only, image-only, and combined image-text settings, and report a visual grounding experiment in which Qwen2VL localizes evidence with near-zero mean IoU. The stated contributions are the new bilingual handwritten benchmark, multi-modality baselines, and an evidence-grounding evaluation.

Significance. If the dataset and annotations were reliable, HW-MLVQA would address a genuine gap: no existing handwritten VQA benchmark combines multilingual scripts with evidence grounding. The paper's inclusion of both commercial and open-source OCR, the ANLS metric, and the explicit grounding evaluation are useful design choices. However, the benchmark's validity is not currently established: the paper gives contradictory dataset statistics, reports OCR word accuracies too low to support the automatic annotation pipeline, provides no quantitative verification of the manually corrected boxes, and no release location. The headline grounding result (mean IoU 0.0166 English, 0.0126 Hindi, Table 5) is therefore difficult to interpret. The strength of the paper is its stated problem framing; the evidence needed to support the central contribution is missing.

major comments (4)
  1. [Abstract, Section 3.4, Table 1] The dataset size is reported inconsistently: the abstract says 1,600 pages and 2,400 QA pairs, Section 3.4 says 1,520 images and 2,400 questions while also mentioning 21,600 train/val questions and 12,000 images, and Table 1 lists 2,400 questions per language (4,800 total). These discrepancies change the claimed scale by a factor of two and make it impossible to know the actual test split; please reconcile all numbers and provide page/question counts per split.
  2. [Section 3.2.2, Table 2, Table 5] Ground-truth evidence boxes are produced by OCR extraction and string matching, but Table 2 reports EasyOCR word accuracy of only 5.56% (English) and 8.29% (Hindi). Unless the manual LabelIMG verification corrected essentially every box, most answer words would not be recovered by the automatic pipeline. No inter-annotator agreement, correction counts, or post-verification accuracy are reported. Consequently, the near-zero mean IoU in Table 5 may reflect incorrect ground-truth boxes rather than poor model grounding, and the grounding claim in Section 5.2.6 is not yet established. Please report verification statistics and validate a random sample with independent annotators.
  3. [Section 5.2.3, Table 4] The image-only English results in the text (EM 45.39, F1 60.13, ANLS 66) do not match Table 4 (EM 57.51, F1 71.32, ANLS 69.11). The value 22.70 that appears in the same sentence is the Hindi image-only EM from Table 4. This inconsistency affects the paper's comparison between modalities and must be corrected.
  4. [Section 3.2, Section 3.4] Because the QA pairs are copied from public SQuAD/MLQA contexts, models pre-trained on those texts may benefit from memorized answers, especially in the text and text+image conditions. The paper does not provide a contamination analysis or new validation questions. In addition, no URL or access information for the dataset is given, so the benchmark cannot be independently assessed or used. Please provide a data release plan and either show that test content is not in pretraining corpora or add a subset of novel handwritten questions.
minor comments (4)
  1. [Section 1, Section 5] The contribution list mentions M-BERT as an evaluated model, but no M-BERT results appear anywhere in the experiments; please either add the results or remove the mention.
  2. [Equation (4)] Equation (4) defines IoU over sets of boxes but does not specify how predicted boxes are matched to ground-truth boxes or whether the value is averaged over all evidence regions per question; please clarify the protocol.
  3. [Section 5.2.6, Figures 10-13] The text states that Fig. 12 illustrates the highest Hindi IoU and then also uses Fig. 12 as a low-IoU example; the narrative and figure captions need to be reconciled.
  4. [Throughout] The manuscript needs careful proofreading: examples include 'we facilitates' in the abstract, 'an sample' in Section 3.4, and inconsistent hyphenation of Qwen2VL/Qwen2-VL and HW-MLVQA/HW-ML VQA.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: HW-MLVQA's QA content is externally sourced from SQuAD/MLQA and model evaluations use external models, so the empirical results are not fitted to the benchmark's own inputs.

full rationale

The derivation chain in HW-MLVQA does not contain a load-bearing circular reduction. The QA pairs are inherited from external corpora (SQuAD [29] and MLQA [12]), not generated from the models under evaluation, and no model parameter is fitted to the benchmark and then re-reported as a prediction. The headline results in Tables 3-5 are measured on external models (LLaMA 3.1, Qwen2VL) and external OCR systems (GoogleOCR, EasyOCR), so the outputs are not defined by the benchmark construction. The one loop that might be suspected is the evidence-box annotation pipeline in Section 3.2.2, where OCR-extracted words are aligned to answer words to produce ground-truth boxes, and Section 5.2.6 then measures model IoU against those boxes. This is a potential annotation-reliability confound rather than a circularity: the boxes are not derived from the evaluated model, and the paper itself reports the low OCR word accuracy (Table 2) that could affect box quality. The self-citations ([7], [8], [15], [19]) are background literature, prior dataset descriptions, or metric sources; none is invoked as an unverified uniqueness theorem or as justification for the central empirical claims. Consequently, no specific circular step can be quoted, and the paper's main evaluation is self-contained with respect to circularity, though the grounding result carries an independent validity risk due to OCR-based ground-truth generation.

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

The benchmark inherits no fitted parameters and introduces no new physical or conceptual entities. It relies on several assumptions: SQuAD/MLQA contexts can be chunked into 50-word pages without losing answerability; OCR-based automatic annotation is accurate enough to produce answer boxes; baseline performance is not inflated by leaked source QA texts; and the handwritten pages are representative of real-world handwriting.

assumptions (5)
  • domain assumption Each SQuAD/MLQA context can be segmented into 50-word chunks without losing the ability to answer the associated questions from a single page.
    Section 3.2 states contexts were divided into 50-word sections; questions from SQuAD/MLQA sometimes require cross-sentence reasoning, and short chunks may make some questions unanswerable from the page alone.
  • domain assumption OCR-based matching of answer words to transcripts produces correct bounding boxes after manual verification.
    Section 3.2.2 describes automatic annotation via Google OCR and EasyOCR plus heuristic alignment; given EasyOCR word accuracy of 5.56% (Table 2), many answer spans may be missed or misaligned before manual checking.
  • domain assumption Baseline scores reflect handling of handwritten inputs rather than memorization of publicly available SQuAD/MLQA texts.
    Section 3.2.1 reuses QA pairs from SQuAD and MLQA; no contamination analysis is provided, yet Qwen2VL hits 67% EM on English ground truth text, which could come from pretraining memorization.
  • standard math Levenshtein distance and set-based IoU, as defined in Equations 3 and 4, are appropriate metrics for answer and evidence evaluation.
    Equations in Section 3.1.2; these are standard, non-controversial metrics.
  • domain assumption The handwritten pages collected from volunteers with black or blue ink on A4 paper are representative of real-world multilingual handwriting.
    Section 3.2.1 describes the collection protocol; representativeness is asserted but not measured.

how reviews work

0 comments
Cite this review

Pith. "Pith review of HW-MLVQA: Elucidating Multilingual Handwritten Document Understanding with a Comprehensive VQA Benchmark." pith.science (2026). https://pith.science/paper/RGWNUCS5

@misc{pith2026250715655,
  author       = {Pith},
  title        = {Pith review of: HW-MLVQA: Elucidating Multilingual Handwritten Document Understanding with a Comprehensive VQA Benchmark},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RGWNUCS5}},
  note         = {Machine review of arXiv:2507.15655}
}
read the original abstract

The proliferation of MultiLingual Visual Question Answering (MLVQA) benchmarks augments the capabilities of large language models (LLMs) and multi-modal LLMs, thereby enabling them to adeptly capture the intricate linguistic subtleties and visual complexities inherent across diverse languages. Despite its potential, the current MLVQA model struggles to fully utilize its capabilities when dealing with the extensive variety of handwritten documents. This article delineates HW-MLVQA, an avant-garde VQA benchmark meticulously crafted to mitigate the dearth of authentic Multilingual Handwritten document comprehension. HW-MLVQA encompasses an extensive collection of 1,600 handwritten Pages complemented by 2,400 question-answers. Furthermore, it provides a robust benchmark evaluation framework spanning three distinct modalities: text, image, and an integrated image & text modality. To simulate authentic real-world contexts devoid of ground truth textual transcriptions, we facilitates a rigorous assessment of proprietary and open-source OCR models. The benchmark aspires to facilitate pivotal advancements in multilingual handwritten document interpretation, fostering innovation and scholarly inquiry within this specialized domain.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 19 canonical work pages

  1. [1]

    In: 2015 IEEE InternationalConferenceonComputerVision (ICCV), pp

    Antol, S., Agrawal, A., Lu, J., Mitchell, M., Batra, D., Zitnick, C.L., Parikh, D.: Vqa: Visual question answering. In: 2015 IEEE InternationalConferenceonComputerVision (ICCV), pp. 2425–2433 (2015). https://doi. org/10.1109/ICCV.2015.279 1

    doi:10.1109/iccv.2015.279 2015

  2. [2]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 15 (CVPR) (2017) 1

    Goyal, Y., Khot, T., Summers-Stay, D., Batra, D., Parikh, D.: Making the v in vqa matter: Elevating the role of image understanding in visual question answering. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 15 (CVPR) (2017) 1

    work page 2017

  3. [3]

    Changpinyo, S., Xue, L., Yarom, M., Thap- liyal, A.V., Szpektor, I., Amelot, J., Chen, X., Soricut, R.: Maxm: Towards multi- lingual visual question answering (2023) arXiv:2209.05401 [cs.CL] 2

    arXiv 2023

  4. [4]

    In: Muresan, S., Nakov, P., Villavicencio, A

    Pfeiffer, J., Geigle, G., Kamath, A., Steitz, J.-M.O., Roth, S., Vulić, I., Gurevych, I.: xGQA: Cross-lingual visual question answer- ing. In: Muresan, S., Nakov, P., Villavicencio, A. (eds.) Findings of the Association for Computational Linguistics: ACL 2022, pp. 2497–2511. Association for Computational Linguistics, Dublin, Ireland (2022). https:// doi.o...

    doi:10.18653/v1/2022.findings-acl.196 2022

  5. [5]

    Vo, D., Tran, K.Q., Nguyen, K.V.: Evjvqa challenge: Multilingual visual ques- tion answering

    Luu-Thuy Nguyen, N., Nguyen, N.H., T.D. Vo, D., Tran, K.Q., Nguyen, K.V.: Evjvqa challenge: Multilingual visual ques- tion answering. Journal of Computer Science and Cybernetics, 237–258 (2023) https://doi.org/10.15625/1813-9663/18157 2

    work page doi:10.15625/1813-9663/18157 2023

  6. [6]

    In: Wong, K.-F., Knight, K., Wu, H

    Gupta, D., Lenka, P., Ekbal, A., Bhat- tacharyya, P.: A unified framework for mul- tilingual and code-mixed visual question answering. In: Wong, K.-F., Knight, K., Wu, H. (eds.) Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 10th International Joint Conference on Natural Language Proc...

    work page 2020

  7. [7]

    IJDAR24(3), 235–249 (2021) 2

    Mathew, M., Gomez, L., Karatzas, D., Jawa- har, C.V.: Asking questions on handwritten document collections. IJDAR24(3), 235–249 (2021) 2

    work page 2021

  8. [8]

    In: Barney Smith, E.H., Liwicki, M., Peng, L

    Mondal, A., Mahadevan, V., Manmatha, R., Jawahar, C.V.: Icdar 2024 competition on recognition and vqa on handwritten doc- uments. In: Barney Smith, E.H., Liwicki, M., Peng, L. (eds.) Document Analysis and Recognition - ICDAR 2024, pp. 426–442. Springer, Cham (2024) 2

    work page 2024

Show all 32 references
  1. [9]

    https: //arxiv.org/abs/2407.21783 2, 8

    The Llama 3 Herd of Models (2024). https: //arxiv.org/abs/2407.21783 2, 8

    2024 arXiv

  2. [10]

    Devlin, J., Chang, M.-W., Lee, K., Toutanova, K.: BERT: Pre-training of deep bidirectional transformers for lan- guage understanding, 4171–4186 (2019) https://doi.org/10.18653/v1/N19-1423 2

    2019 doi

  3. [11]

    arXiv preprint arXiv:2409.12191 (2024) 2, 8

    Wang, P., Bai, S., Tan, S., Wang, S., Fan, Z., Bai, J., Chen, K., Liu, X., Wang, J., Ge, W., Fan, Y., Dang, K., Du, M., Ren, X., Men, R., Liu, D., Zhou, C., Zhou, J., Lin, J.: Qwen2-vl: Enhancing vision-language model’s percep- tion of the world at any resolution. arXiv prepri...

    2024 arXiv

  4. [12]

    In: Jurafsky, D., Chai, J., Schluter, N., Tetreault, J

    Lewis, P., Oguz, B., Rinott, R., Riedel, S., Schwenk, H.: MLQA: Evaluating cross-lingual extractive question answer- ing. In: Jurafsky, D., Chai, J., Schluter, N., Tetreault, J. (eds.) Proceedings of the 58th Annual Meeting of the Asso- ciation for Computational Linguistics, p...

    2020 doi

  5. [13]

    Transactions of the Asso- ciation for Computational Linguistics8, 454– 470 (2020) https://doi.org/10.1162/tacl_a_ 00317 2

    Clark, J.H., Choi, E., Collins, M., Garrette, D., Kwiatkowski, T., Nikolaev, V., Palomaki, J.: TyDi QA: A benchmark for information- seeking question answering in typologically diverse languages. Transactions of the Asso- ciation for Computational Linguistics8, 454– 470 (2020)...

    2020 doi

  6. [14]

    In: 2019 IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (CVPR), pp

    Singh, A., Natarajan, V., Shah, M., Jiang, Y., Chen, X., Batra, D., Parikh, D., Rohrbach, M.: TextVQA: Towards VQA Models That Can Read . In: 2019 IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (CVPR), pp. 8309–8318. IEEE Computer Society, Los Alamitos, CA, U...

    2019 doi

  7. [15]

    In: 2021 IEEE Winter Conference on Applications of Computer Vision (WACV), 16 pp

    Mathew, M., Karatzas, D., Jawahar, C.V.: Docvqa: A dataset for vqa on document images. In: 2021 IEEE Winter Conference on Applications of Computer Vision (WACV), 16 pp. 2199–2208 (2021). https://doi.org/10. 1109/WACV48630.2021.00225 2, 3, 9

    2021

  8. [16]

    Tang, J., Liu, Q., Ye, Y., Lu, J., Wei, S., Lin, C., Li, W., Mahmood, M.F.F.B., Feng, H., Zhao, Z., Wang, Y., Liu, Y., Liu, H., Bai, X., Huang, C.: MTVQA: Benchmarking Multilingual Text-Centric Visual Question Answering (2024) 2

    2024

  9. [17]

    International Jour- nal on Document Analysis and Recognition 5(1), 39–46 (2002) https://doi.org/10.1007/ s100320200071 2

    Marti, U.-V., Bunke, H.: The iam-database: an english sentence database for offline handwriting recognition. International Jour- nal on Document Analysis and Recognition 5(1), 39–46 (2002) https://doi.org/10.1007/ s100320200071 2

    2002

  10. [18]

    In: International Conference on Document AnalysisandRecognition,pp.399–412(2021) 2

    Lee, A.W., Chung, J., Lee, M.: GNHK: a dataset for english handwriting in the wild. In: International Conference on Document AnalysisandRecognition,pp.399–412(2021) 2

    2021

  11. [19]

    In: Interna- tional Conference on Document Analysis and Recognition, pp

    Mondal, A., Tulsyan, K., Jawahar, C.: Bridg- ing the gap in resource for offline english handwritten text recognition. In: Interna- tional Conference on Document Analysis and Recognition, pp. 413–428 (2024) 2

    2024

  12. [20]

    In: 2011 International Conference on Document Analysis and Recognition, pp

    Grosicki, E., El-Abed, H.: Icdar 2011-french handwriting recognition competition. In: 2011 International Conference on Document Analysis and Recognition, pp. 1459–1463 (2011) 2

    2011

  13. [21]

    In: 2011 International Confer- ence on Document Analysis and Recognition, pp

    Liu, C.-L., Yin, F., Wang, D.-H., Wang, Q.- F.: Casia online and offline chinese handwrit- ing databases. In: 2011 International Confer- ence on Document Analysis and Recognition, pp. 37–41 (2011). https://doi.org/10.1109/ ICDAR.2011.17 2

    2011

  14. [22]

    In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp

    Zhu, Y., Groth, O., Bernstein, M., Fei-Fei, L.: Visual7W: Grounded Question Answer- ing in Images . In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 4995–5004. IEEE Computer Society, Los Alamitos, CA, USA (2016). https://doi.org/10.1109/CVPR.2016.540 2

    2016 doi

  15. [23]

    In: 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp

    Hudson, D.A., Manning, C.D.: GQA: A New Dataset for Real-World Visual Reasoning and Compositional Question Answering . In: 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 6693–

    2019

  16. [24]

    Das, A., Agrawal, H., Zitnick, C.L., Parikh, D., Batra, D.: Human Attention in Visual Question Answering: Do Humans and Deep Networks Look at the Same Regions? In: Conference on Empirical Methods in Natural Language Processing (EMNLP) (2016) 2

    2016

  17. [25]

    In: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp

    Johnson, J., Hariharan, B., Maaten, L., Fei- Fei, L., Zitnick, C.L., Girshick, R.: CLEVR: A Diagnostic Dataset for Compositional Lan- guage and Elementary Visual Reasoning . In: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1988–

    2017

  18. [26]

    In: Moschitti, A., Pang, B., Daele- mans, W

    Kazemzadeh, S., Ordonez, V., Matten, M., Berg, T.: ReferItGame: Referring to objects in photographs of natural scenes. In: Moschitti, A., Pang, B., Daele- mans, W. (eds.) Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 787–79...

    2014 doi

  19. [27]

    In: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp

    Gurari, D., Li, Q., Stangl, A.J., Guo, A., Lin, C., Grauman, K., Luo, J., Bigham, J.P.: VizWiz Grand Challenge: Answering Visual Questions from Blind People . In: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 3608–

    2018

  20. [28]

    IJCV (2020) 3

    Kuznetsova, A., Rom, H., Alldrin, N., Uijlings, J., Krasin, I., Pont-Tuset, J., Kamali, S., Popov, S., Malloci, M., Kolesnikov, A., Duerig, T., Ferrari, V.: The open images dataset v4: Unified image 17 classification, object detection, and visual relationship detection at scal...

    2020

  21. [29]

    In: Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing (2016) 4 18

    Rajpurkar, P., Zhang, J., Lopyrev, K., Liang, P.: Squad: 100,000+ questions for machine comprehension of text. In: Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing (2016) 4 18

    2016

  22. [1997]

    https://doi.org/10.1109/ CVPR.2017.215 2

    IEEE Computer Society, Los Alamitos, CA, USA (2017). https://doi.org/10.1109/ CVPR.2017.215 2

    2017

  23. [3617]

    https://doi.org/10.1109/ CVPR.2018.00380 3

    IEEE Computer Society, Los Alamitos, CA, USA (2018). https://doi.org/10.1109/ CVPR.2018.00380 3

    2018

  24. [6702]

    https://doi.org/10.1109/ CVPR.2019.00686 2

    IEEE Computer Society, Los Alamitos, CA, USA (2019). https://doi.org/10.1109/ CVPR.2019.00686 2

    2019

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.