UIHash: Detecting Similar Android UIs through Grid-Based Visual Appearance Representation

Authors: 

Jiawei Li, Beihang University; National University of Singapore; Jian Mao, Beihang University; Tianmushan Laboratory; Hangzhou Innovation Institute, Beihang University; Jun Zeng, National University of Singapore; Qixiao Lin and Shaowen Feng, Beihang University; Zhenkai Liang, National University of Singapore

Abstract: 

User interfaces (UIs) is the main channel for users to interact with mobile apps. As such, attackers often create similar-looking UIs to deceive users, causing various security problems, such as spoofing and phishing. Prior studies identify these similar UIs based on their layout trees or screenshot images. These techniques, however, are susceptible to being evaded. Guided by how users perceive UIs and the features they prioritize, we design a novel grid-based UI representation to capture UI visual appearance while maintaining robustness against evasion. We develop an approach, UIHash, to detect similar Android UIs by comparing their visual appearance. It divides the UI into a #-shaped grid and abstracts UI controls across screen regions, then calculates UI similarity through a neural network architecture that includes a convolutional neural network and a Siamese network. Our evaluation shows that UIHash achieves an F1-score of 0.984 in detection, outperforming existing tree-based methods and image-based methods. Moreover, we have discovered evasion techniques that circumvent existing detection approaches.

Open Access Media

USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.