Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -21,4 +21,82 @@ dataset_info:
|
|
21 |
num_examples: 7997
|
22 |
download_size: 1178990085
|
23 |
dataset_size: 1258281789.658
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
num_examples: 7997
|
22 |
download_size: 1178990085
|
23 |
dataset_size: 1258281789.658
|
24 |
+
task_categories:
|
25 |
+
- object-detection
|
26 |
+
tags:
|
27 |
+
- ui
|
28 |
+
- design
|
29 |
+
- detection
|
30 |
+
size_categories:
|
31 |
+
- n<1K
|
32 |
---
|
33 |
+
|
34 |
+
# Dataset: Mobile UI Design Detection
|
35 |
+
|
36 |
+
## Introduction
|
37 |
+
|
38 |
+
This dataset is designed for object detection tasks with a focus on detecting elements in mobile UI designs. The targeted objects include text, images, and groups. The dataset contains images and object detection boxes, including class labels and location information.
|
39 |
+
|
40 |
+
## Dataset Content
|
41 |
+
|
42 |
+
Load the dataset and take a look at an example:
|
43 |
+
|
44 |
+
```python
|
45 |
+
>>> from datasets import load_dataset
|
46 |
+
>>>> ds = load_dataset("mrtoy/mobile-ui-design")
|
47 |
+
>>> example = ds[0]
|
48 |
+
>>> example
|
49 |
+
{'width': 375,
|
50 |
+
'height': 667,
|
51 |
+
'image': <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=375x667>,
|
52 |
+
'objects': {'bbox': [[0.0, 0.0, 375.0, 667.0],
|
53 |
+
[0.0, 0.0, 375.0, 667.0],
|
54 |
+
[0.0, 0.0, 375.0, 20.0],
|
55 |
+
...
|
56 |
+
],
|
57 |
+
'category': ['artboard',
|
58 |
+
'rectangle',
|
59 |
+
'rectangle',
|
60 |
+
...]}}
|
61 |
+
```
|
62 |
+
|
63 |
+
The dataset has the following fields:
|
64 |
+
|
65 |
+
- image: PIL.Image.Image object containing the image.
|
66 |
+
- height: The image height.
|
67 |
+
- width: The image width.
|
68 |
+
- objects: A dictionary containing bounding box metadata for the objects in the image:
|
69 |
+
- bbox: The object’s bounding box (xmin,ymin,width,height).
|
70 |
+
- category: The object’s category, with possible values including artboard、rectangle、text、group、...
|
71 |
+
|
72 |
+
You can visualize the bboxes on the image using some internal torch utilities.
|
73 |
+
|
74 |
+
```python
|
75 |
+
import torch
|
76 |
+
from torchvision.ops import box_convert
|
77 |
+
from torchvision.utils import draw_bounding_boxes
|
78 |
+
from torchvision.transforms.functional import pil_to_tensor, to_pil_image
|
79 |
+
|
80 |
+
item = ds[0]
|
81 |
+
boxes_xywh = torch.tensor(item['objects']['bbox'])
|
82 |
+
boxes_xyxy = box_convert(boxes_xywh, 'xywh', 'xyxy')
|
83 |
+
to_pil_image(
|
84 |
+
draw_bounding_boxes(
|
85 |
+
pil_to_tensor(item['image']),
|
86 |
+
boxes_xyxy,
|
87 |
+
labels=item['objects']['category'],
|
88 |
+
)
|
89 |
+
)
|
90 |
+
```
|
91 |
+
|
92 |
+
|
93 |
+
## Applications
|
94 |
+
|
95 |
+
This dataset can be used for various applications, such as:
|
96 |
+
|
97 |
+
- Training and evaluating object detection models for mobile UI designs.
|
98 |
+
- Identifying design patterns and trends to aid UI designers and developers in creating high-quality mobile app UIs.
|
99 |
+
- Enhancing the automation process in generating UI design templates.
|
100 |
+
- Improving image recognition and analysis in the field of mobile UI design.
|
101 |
+
|
102 |
+
|