2022
DOI: 10.1109/tvcg.2021.3070876
|View full text |Cite
|
Sign up to set email alerts
|

Deep Colormap Extraction From Visualizations

Abstract: This work presents a new approach based on deep learning to automatically extract colormaps from visualizations. After summarizing colors in an input visualization image as a Lab color histogram, we pass the histogram to a pre-trained deep neural network, which learns to predict the colormap that produces the visualization. To train the network, we create a new dataset of ∼64K visualizations that cover a wide variety of data distributions, chart types, and colormaps. The network adopts an atrous spatial pyrami… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
4
0

Year Published

2022
2022
2024
2024

Publication Types

Select...
6
3

Relationship

3
6

Authors

Journals

citations
Cited by 19 publications
(4 citation statements)
references
References 57 publications
0
4
0
Order By: Relevance
“…Therefore, AIassisted tools have leveraged existing examples to guide colormap construction. For example, deep learning can enable colormap style transfer from existing images to new charts [37,64,76,77]. However, users may not have quality example images.…”
Section: Tools For Designing Sequential Colormapsmentioning
confidence: 99%
“…Therefore, AIassisted tools have leveraged existing examples to guide colormap construction. For example, deep learning can enable colormap style transfer from existing images to new charts [37,64,76,77]. However, users may not have quality example images.…”
Section: Tools For Designing Sequential Colormapsmentioning
confidence: 99%
“…Despite the importance, the design space of chart attributes is too large for designers to choose from, resulting in various types of charts in designers' own style [SLC * 23]. Most of these works follow a coarse‐to‐fine strategy to first identify chart types and then extract visual marks and channels [SKC * 11,JKS * 17,PMH17,YZF * 22].…”
Section: Related Workmentioning
confidence: 99%
“…Third, it can facilitate the construction of large-scale visualization corpora. Most of the existing visualization corpora only provide high-level labels of visualizations such as the visualization type [18] and color usage [55]. With our approach, fine-grained labels (e.g., the trend of data) of visualizations can be easily labeled based on similar visualization retrieval.…”
Section: Generalizability and Application Scenariosmentioning
confidence: 99%