VIS30K: A Collection of Figures and Tables From IEEE Visualization Conference Publications

Chen, Jian; Ling, Meng; Li, Rui; Isenberg, Petra; Isenberg, Tobias; Sedlmair, Michael; Möller, Torsten B.; Laramee, Robert S.; Shen, Han‐Wei; Katharina, Wünsche,; Wang, Qiru

doi:10.1109/tvcg.2021.3054916

Cited by 40 publications

(29 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Zeng et al [31] contributed VIStory, a technique for exploring the figures in VIS publications. Chen et al [32] adopted object detection models to extract the figures and tables in IEEE VIS publications and proposed VIS30K. These studies mainly focus on perception tasks and analytical techniques for figures, instead of the visual designs inside.…”

Section: Meta Analysis On Visualization Publicationsmentioning

confidence: 99%

Revisiting the Design Patterns of Composite Visualizations

Deng¹,

Cui²,

Meng³

et al. 2022

Preprint

View full text Add to dashboard Cite

Composite visualization is a popular design strategy that represents complex datasets by integrating multiple visualizations in a meaningful and aesthetic layout, such as juxtaposition, overlay, and nesting. With this strategy, numerous novel designs have been proposed in visualization publications to accomplish various visual analytic tasks. These well-crafted composite visualizations have formed a valuable collection for designers and researchers to address real-world problems and inspire new research topics and designs. However, there is a lack of understanding of design patterns of composite visualization, thus failing to provide holistic design space and concrete examples for practical use. In this paper, we opted to revisit the composite visualizations in VIS publications and answered what and how visualizations of different types are composed together. To achieve this, we first constructed a corpus of composite visualizations from IEEE VIS publications and decomposed them into a series of basic visualization types (e.g., bar chart, map, and matrix). With this corpus, we studied the spatial (e.g., separated or overlaying) and semantic relationships (e.g., with same types or shared axis) between visualizations and proposed a taxonomy consisting of eight different design patterns (e.g., repeated, stacked, accompanied, and nested). Furthermore, we analyzed and discussed common practices of composite visualizations, such as the distribution of different patterns and correlations between visualization types. From the analysis and examples, we obtained insights into different design patterns on the utilities, advantages, and disadvantages. Finally, we developed an interactive system to help visualization developers and researchers conveniently explore collected examples and design patterns.

show abstract

Section: Meta Analysis On Visualization Publicationsmentioning

confidence: 99%

Revisiting the Design Patterns of Composite Visualizations

Deng¹,

Cui²,

Meng³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…• Algorithms, figures, tables, and equations. In the examples in this paper, we rely on the VIS30K dataset [8,9] for this purpose. • Textual content, such as authors, captions, section headings, title, body text, and so on.…”

Section: Document Domain Randomizationmentioning

confidence: 99%

“…We did not use these sources since each of them covers only a single facet of the rich scholarly article genre and, since these images are often modern, they do not contain images from scanned documents and thus could potentially bias CNN's classification accuracy. Here, we chose VIS30K [8,9], a comprehensive collection of images including tables, figures, algorithms, and equations. The figures in VIS30K contain not only charts and tables but also spatial data and photos.…”

Section: Document Domain Randomizationmentioning

confidence: 99%

“…CS-150 had diverse styles collected from several computer science conferences. Two additional domain-specific sets were chosen based on our own interests and familiarity: ACL300 had 300 randomly sampled articles (or 2508 pages) from the 55,759 papers scraped from the ACL anthology website; VIS300 contains about 10% (or 2619 pages) of the document pages in randomly partitioned articles from 26,350 VIS paper pages of the past 30 years in Chen et al [9]. Using these two specialized domains lets us test H2 to measure the effect of using images generated in one domain to test on another when the reality gap could be large.…”

Section: Evaluation Of Ddrmentioning

confidence: 99%

“…The VIS30K [9] dataset comprises all the images published at IEEE visualization conferences in each year, rather than just a few samples. All image files are copyrighted and for most the copyright is owned by IEEE.…”

Section: F Image Rights and Attributionmentioning

confidence: 99%

See 2 more Smart Citations

Document Domain Randomization for Deep Learning Document Layout Extraction

Ling,

Chen,

Möller

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

We present document domain randomization (DDR), the first successful transfer of CNNs trained only on graphically rendered pseudo-paper pages to real-world document segmentation. DDR renders pseudo-document pages by modeling randomized textual and non-textual contents of interest, with userdefined layout and font styles to support joint learning of fine-grained classes. We demonstrate competitive results using our DDR approach to extract nine document classes from the benchmark CS-150 and papers published in two domains, namely annual meetings of Association for Computational Linguistics (ACL) and IEEE Visualization (VIS). We compare DDR to conditions of style mismatch, fewer or more noisy samples that are more easily obtained in the real world. We show that high-fidelity semantic information is not necessary to label semantic classes but style mismatch between train and test can lower model accuracy. Using smaller training samples had a slightly detrimental effect. Finally, network models still achieved high test accuracy when correct labels are diluted towards confusing labels; this behavior hold across several classes.

show abstract