Datasets:
v4.0.0 release notes
#2
by
kyamagu
- opened
README.md
CHANGED
@@ -648,6 +648,22 @@ configs:
|
|
648 |
|
649 |
The Crello dataset is compiled for the study of vector graphic documents. The dataset contains document meta-data such as canvas size and pre-rendered elements such as images or text boxes. The original templates were collected from [crello.com](https://crello.com) (now [create.vista.com](https://create.vista.com/)) and converted to a low-resolution format suitable for machine learning analysis.
|
650 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
651 |
### Supported Tasks and Leaderboards
|
652 |
|
653 |
[CanvasVAE](https://arxiv.org/abs/2108.01249) studies unsupervised document generation.
|
@@ -773,18 +789,20 @@ In the following, categorical fields are shown as `categorical` type, but the ac
|
|
773 |
|
774 |
**Canvas attributes**
|
775 |
|
776 |
-
| Field | Type | Shape | Description
|
777 |
-
| ------------- | ----------- | ------- |
|
778 |
-
| id | string | () | Template ID from crello.com
|
779 |
-
| group | categorical | () | Broad design groups, such as social media posts or blog headers
|
780 |
-
| format | categorical | () | Detailed design formats, such as Instagram post or postcard
|
781 |
-
| category | categorical | () | Topic category of the design, such as holiday celebration
|
782 |
-
| canvas_width | categorical | () | Canvas pixel width
|
783 |
-
| canvas_height | categorical | () | Canvas pixel height
|
784 |
-
| length | int64 | () | Length of elements
|
785 |
-
| suitability | categorical | (None,) | List of display tags, only `mobile` tag exists
|
786 |
-
| keywords | string | (None,) | List of keywords associated to this template
|
787 |
-
| industries | categorical | (None,) | List of industry tags like `marketingAds`
|
|
|
|
|
788 |
|
789 |
**Element attributes**
|
790 |
|
@@ -817,13 +835,13 @@ https://create.vista.com/artboard/?template=<template_id>
|
|
817 |
|
818 |
### Data Splits
|
819 |
|
820 |
-
The Crello dataset has 3 splits: train, validation, and test. The current split is generated
|
821 |
|
822 |
| Split | Count |
|
823 |
| --------- | ----- |
|
824 |
-
| train |
|
825 |
-
| validaton |
|
826 |
-
| test |
|
827 |
|
828 |
|
829 |
### Visualization
|
@@ -941,9 +959,15 @@ We do not re-distribute the original files as we are not allowed by terms.
|
|
941 |
|
942 |
### Releases
|
943 |
|
|
|
|
|
|
|
|
|
|
|
|
|
944 |
3.1: bugfix release (Feb 16, 2023)
|
945 |
|
946 |
-
- Fix a bug that ignores newline characters in some of the texts
|
947 |
|
948 |
3.0: v3 release (Feb 13, 2023)
|
949 |
|
|
|
648 |
|
649 |
The Crello dataset is compiled for the study of vector graphic documents. The dataset contains document meta-data such as canvas size and pre-rendered elements such as images or text boxes. The original templates were collected from [crello.com](https://crello.com) (now [create.vista.com](https://create.vista.com/)) and converted to a low-resolution format suitable for machine learning analysis.
|
650 |
|
651 |
+
### Usage
|
652 |
+
|
653 |
+
```python
|
654 |
+
import datasets
|
655 |
+
|
656 |
+
dataset = datasets.load_dataset("cyberagent/crello")
|
657 |
+
```
|
658 |
+
|
659 |
+
Old revision is available via `revision` option.
|
660 |
+
|
661 |
+
```python
|
662 |
+
import datasets
|
663 |
+
|
664 |
+
dataset = datasets.load_dataset("cyberagent/crello", revision="3.1")
|
665 |
+
```
|
666 |
+
|
667 |
### Supported Tasks and Leaderboards
|
668 |
|
669 |
[CanvasVAE](https://arxiv.org/abs/2108.01249) studies unsupervised document generation.
|
|
|
789 |
|
790 |
**Canvas attributes**
|
791 |
|
792 |
+
| Field | Type | Shape | Description |
|
793 |
+
| ------------- | ----------- | ------- | ----------------------------------------------------------------- |
|
794 |
+
| id | string | () | Template ID from crello.com |
|
795 |
+
| group | categorical | () | Broad design groups, such as social media posts or blog headers |
|
796 |
+
| format | categorical | () | Detailed design formats, such as Instagram post or postcard |
|
797 |
+
| category | categorical | () | Topic category of the design, such as holiday celebration |
|
798 |
+
| canvas_width | categorical | () | Canvas pixel width |
|
799 |
+
| canvas_height | categorical | () | Canvas pixel height |
|
800 |
+
| length | int64 | () | Length of elements |
|
801 |
+
| suitability | categorical | (None,) | List of display tags, only `mobile` tag exists |
|
802 |
+
| keywords | string | (None,) | List of keywords associated to this template |
|
803 |
+
| industries | categorical | (None,) | List of industry tags like `marketingAds` |
|
804 |
+
| preview | image | () | Preview image of the template for convenience; only for debugging |
|
805 |
+
| cluster_index | int64 | () | Cluster index used to split the dataset; only for debugging |
|
806 |
|
807 |
**Element attributes**
|
808 |
|
|
|
835 |
|
836 |
### Data Splits
|
837 |
|
838 |
+
The Crello dataset has 3 splits: train, validation, and test. The current split is generated based on appearance-based clustering.
|
839 |
|
840 |
| Split | Count |
|
841 |
| --------- | ----- |
|
842 |
+
| train | 19095 |
|
843 |
+
| validaton | 1951 |
|
844 |
+
| test | 2375 |
|
845 |
|
846 |
|
847 |
### Visualization
|
|
|
959 |
|
960 |
### Releases
|
961 |
|
962 |
+
4.0.0: v4 release (Dec 5, 2023)
|
963 |
+
|
964 |
+
- Change the dataset split based on the template appearance to avoid near-duplicates: no compatibility with v3.
|
965 |
+
- Class labels have been reordered: no compabilitity with v3.
|
966 |
+
- Small improvement to font rendering.
|
967 |
+
|
968 |
3.1: bugfix release (Feb 16, 2023)
|
969 |
|
970 |
+
- Fix a bug that ignores newline characters in some of the texts.
|
971 |
|
972 |
3.0: v3 release (Feb 13, 2023)
|
973 |
|