Datasets:
ArXiv:
License:
Update README.md
Browse files
README.md
CHANGED
|
@@ -10,7 +10,7 @@ Each value is a [0, 1] decimal number indicating the ratio of the corresponding
|
|
| 10 |
|
| 11 |
The data we released is divided into three domains: mobile, desktop and web.
|
| 12 |
|
| 13 |
-
All annotation data
|
| 14 |
* `img_filename`: the interface screenshot file
|
| 15 |
* `instruction`: human instruction
|
| 16 |
* `bbox`: the bounding box of the target element corresponding to instruction
|
|
@@ -19,12 +19,12 @@ Some data also contains a `data_type`, which records the type of an element in i
|
|
| 19 |
|
| 20 |
### Mobile data
|
| 21 |
|
| 22 |
-
|
| 23 |
|
| 24 |
#### AMEX
|
| 25 |
Android Multi-annotation EXpo (AMEX) is a comprehensive, large-scale dataset designed for generalist mobile GUI-control agents [1].
|
| 26 |
|
| 27 |
-
The annotation data
|
| 28 |
|
| 29 |
-'amex_raw.json'
|
| 30 |
|
|
@@ -45,7 +45,7 @@ cat amex_images_part_* > amex_images.zip
|
|
| 45 |
|
| 46 |
This is a dataset extended from Rico dataset [2] for two tasks: similar UI component retrieval and referring expression component retrieval [3].
|
| 47 |
|
| 48 |
-
The annotation data
|
| 49 |
|
| 50 |
- `uibert_raw.json`
|
| 51 |
|
|
@@ -57,7 +57,7 @@ The UIBert images are stored in
|
|
| 57 |
Widget Captioning data are collected by [4].
|
| 58 |
RICOSCA is a dataset automatically labeled using Android VH in [5]
|
| 59 |
|
| 60 |
-
The annotation data
|
| 61 |
|
| 62 |
- `widget_captioning.json`
|
| 63 |
- `ricosca.json`
|
|
@@ -70,7 +70,7 @@ The rico images are stored in
|
|
| 70 |
|
| 71 |
This part of data are sampled from a android environment for building and benchmarking autonomous computer control agents [6].
|
| 72 |
|
| 73 |
-
The annotation data
|
| 74 |
|
| 75 |
- `aw_mobile.json`
|
| 76 |
|
|
@@ -78,3 +78,9 @@ The rico images are stored in
|
|
| 78 |
|
| 79 |
- `mobile_images.zip`
|
| 80 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
|
| 11 |
The data we released is divided into three domains: mobile, desktop and web.
|
| 12 |
|
| 13 |
+
All annotation data is stored in JSON format and each sample contains:
|
| 14 |
* `img_filename`: the interface screenshot file
|
| 15 |
* `instruction`: human instruction
|
| 16 |
* `bbox`: the bounding box of the target element corresponding to instruction
|
|
|
|
| 19 |
|
| 20 |
### Mobile data
|
| 21 |
|
| 22 |
+
This part of data is stored under the *mobile_domain* directory. Our mobile grounding data consists of four parts.
|
| 23 |
|
| 24 |
#### AMEX
|
| 25 |
Android Multi-annotation EXpo (AMEX) is a comprehensive, large-scale dataset designed for generalist mobile GUI-control agents [1].
|
| 26 |
|
| 27 |
+
The annotation data is stored in
|
| 28 |
|
| 29 |
-'amex_raw.json'
|
| 30 |
|
|
|
|
| 45 |
|
| 46 |
This is a dataset extended from Rico dataset [2] for two tasks: similar UI component retrieval and referring expression component retrieval [3].
|
| 47 |
|
| 48 |
+
The annotation data is stored in
|
| 49 |
|
| 50 |
- `uibert_raw.json`
|
| 51 |
|
|
|
|
| 57 |
Widget Captioning data are collected by [4].
|
| 58 |
RICOSCA is a dataset automatically labeled using Android VH in [5]
|
| 59 |
|
| 60 |
+
The annotation data is stored in
|
| 61 |
|
| 62 |
- `widget_captioning.json`
|
| 63 |
- `ricosca.json`
|
|
|
|
| 70 |
|
| 71 |
This part of data are sampled from a android environment for building and benchmarking autonomous computer control agents [6].
|
| 72 |
|
| 73 |
+
The annotation data is stored in
|
| 74 |
|
| 75 |
- `aw_mobile.json`
|
| 76 |
|
|
|
|
| 78 |
|
| 79 |
- `mobile_images.zip`
|
| 80 |
|
| 81 |
+
### Desktop data
|
| 82 |
+
|
| 83 |
+
This part of data is stored under the *desktop_domain* directory. All of the desktop grounding data is collected from the real environments of personal computers running different operating systems. Each image is split into multiple sub-images to enhance data diversity.
|
| 84 |
+
Our mobile grounding data consists of three parts.
|
| 85 |
+
|
| 86 |
+
|