Official Repo of “VINS: Visual Search for Mobile User Interface Design” published in CHI 2021
VINS dataset is an annotated dataset containing a representative collection of UI screens across two design stages: abstract wireframes and high-fidelity fully designed interfaces. All of these UIs are annotated with bounding boxes spanning different classes of UI components. We identified a total of 11 UI components with varying functionality: background images, sliding menus, pop-up windows, input fields, icons, images, texts, switches, checked views, text buttons, and page indicators. Based on our analysis and due to relatively small training instances, we combined radio buttons and checkboxes to represent the checked view class.
VINS dataset has a total of 4,800 images of UI screens of different design stages to ensure that the VINS can perform on a wider variety of design inputs. It includes the following:
-
257 images of abstract wireframes UIs
-
4,543 images of high-fidelity screens:
- 2000 images of Android screen collected from Rico Dataset
- 740 images of new Android screens
- 1200 images of iphone screens
- 603 images of UI design collected from design sharing websites