Look, Read and Enrich - Learning from Scientific Figures and their Captions
Open accessAbstract
Compared to natural images, understanding scientific figures is particularly hard for machines. However, there is a valuable source of information in scientific literature that until now has remained untapped: the correspondence between a figure and its caption. In this paper we investigate what can be learnt by looking at a large number of figures and reading their captions, and introduce a figure-caption correspondence learning task that makes use of our observations. Training visual and language networks without supervision other than pairs of unconstrained figures and captions is shown to successfully solve this task. We also show that transferring lexical and semantic knowledge from a knowledge graph significantly enriches the resulting features. Finally, we demonstrate the positive impact of such features in other tasks involving scientific text and figures, like multi-modal classification and machine comprehension for question answering, outperforming supervised baselines and ad-hoc approaches.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Writing alt text becomes a nontrivial task without domain expertise, even for experienced accessibility professionals.
“Therefore, writing alt text becomes a nontrivial task without domain expertise, even for experienced accessibility professionals [25, 33, 79].”
From How University Disability Services Professionals Write Image Descriptions for HCI Figures Using Generative AI (2026) · cites this paper as [25]
Related work (p. 1) says scientific figures are abstract and their captions use specialized language; Conclusions (p. 8) describes the difficulty of understanding figures for machines. Neither passage addresses writing alt text without domain expertise or the experience of accessibility professionals. Searches for alt text, accessibility professionals, and human-written descriptions found no passage addressing that claim. (LLM, conf=0.72)
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
Deep Contextualized Word Representations
Peters, M., Neumann, M., Iyyer, M., Gardner, M., Clark, C., Lee, K., & Zettlemoyer, L. (2018) · Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers) · doi:10.18653/v1/n18-1202 · Cited by 11,983 · see citing works
Kembhavi, A., Seo, M., Schwenk, D., Choi, J., Farhadi, A., & Hajishirzi, H. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.571 · Cited by 244 · cited 6× · see citing works
“We leverage the TQA dataset and the baselines in [18] to evaluate the features learnt by the FCC task in a multi-modal machine comprehension scenario.”
BackgroundMethodInfluential
Flagged first, then by citations
Research works printed but not linked: 10
- [Ammar et al. , 2018] W Ammar, D Groeneveld, C Bhagavatula, I Beltagy, M Crawford, D Downey, J Dunkelberger, A Elgohary, S Feldman, V Ha, R Kinney, S Kohlmeier, K Lo, T Murray, HH Ooi, ME Peters, J Power, S Skjonsberg, LL Wang, C Wilhelm, Z Yuan, M van Zuylen, and OEtzioni. Construction of the literature graph in semantic scholar. In NAACL-HTL , 2018.
- [Antol et al. , 2015] S Antol, A Agrawal, J Lu, M Mitchell, D Batra, CL Zitnick, and D Parikh. Vqa: Visual question answering. 2015 IEEE International Conference on Computer Vision (ICCV) , pages 2425-2433, 2015.
- [Arandjelovic and Zisserman, 2017] R Arandjelovic and A Zisserman. Look, listen and learn. 2017 IEEE International Conference on Computer Vision (ICCV) , pages 609-617, 2017.
- [Blum and Mitchell, 1998] A Blum and T Mitchell. Combining labeled and unlabeled data with co-training. In COLT , 1998.
- [Bojanowski et al. , 2017] P Bojanowski, E Grave, A Joulin, and T Mikolov. Enriching word vectors with subword information. Transactions of the Association for Computational Linguistics , 5:135-146, 2017.
- [Clark and Divvala, 2016] C Clark and S Divvala. Pdffigures 2.0: Mining figures from research papers. 2016 IEEE/ACM Joint Conference on Digital Libraries (JCDL) , pages 143-152, 2016.
- [Denaux and Gomez-Perez, 2019] R Denaux and JM Gomez-Perez. Vecsigrafo: Corpus-based word- concept embeddings. Semantic Web , pages 68-96, 2019.
- [Eisenschtat and Wolf, 2017] A Eisenschtat and L Wolf. Linking image and text with 2-way nets. 2017 IEEE Conf. on Computer Vision and Pattern Recognition (CVPR) , 2017.
- [Engilberge et al. , 2018] M Engilberge, L Chevallier, P P´ erez, and M Cord. Finding beans in burgers: Deep semantic-visual embedding with localization. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2018.
- [Faghri et al. , 2017] F Faghri, DJ Fleet, J Kiros, and S Fidler. Vse++: Improving visual-semantic embeddings with hard negatives. In BMVC , 2017.
Cited by
Sorted by citations
Raees, M., Iwamoto, Y., Papangelis, K., Heard, J., & Tigwell, G. W. (2026) · arXiv · doi:10.48550/arxiv.2602.08937 · Cited by 1 · see citing works
“Therefore, writing alt text becomes a nontrivial task without domain expertise, even for experienced accessibility professionals [32, 73, 24].”
- Authors
- Gomez-Perez, J. M., & Ortega, R.
- Affiliations
- Expert System Cogito Labs, Madrid, Spain
- Venue
- Proceedings of the 10th International Conference on Knowledge Capture
- Publisher
- ACM
- Published
- 2019-09-23
- Pages
- 101-108
- OpenAlex
- W2974318400
- Cited by
- 8
- References
- 2
- Citations by year
- 2025: 1 · 2024: 1 · 2023: 3 · 2022: 1 · 2021: 1 · 2020: 1
- FWCI
- 0.24
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- International Conference on Knowledge Capture (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Zenodo (CERN European Organization for Nuclear Research) (submittedVersion, cc-by)
- arXiv (Cornell University)
- Funders
- Horizon 2020 Framework Programme (825627)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Multimodal Machine Learning Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Handwritten Text Recognition Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Image Retrieval and Classification Techniques
- Keywords
- scientific figures, figure captions, unsupervised learning, knowledge graphs, multimodal classification, machine reading comprehension, question answering
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
