This disclosure describes the preservation of important elements of a user interface (UI) during retargeting of the interface image on a mobile device. An on-device machine learned (ML) model is utilized to detect saliency or lack thereof of various UI elements in the user interface. Training of the ML model is performed by utilizing training data from repositories of software application designs and screenshot data from online marketplaces and app evaluation services.

The trained ML model is utilized to detect salient UI elements that are to be preserved during display retargeting. During resizing of the UI, with express user permission, content-aware image retargeting techniques are utilized for the preservation of elements identified as important by the UI saliency detection model. Past interactions are utilized, and interpretation or corrective action is performed only upon permission from the user.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.