No abstract

Deep Neural Networks (DNNs) are susceptible to adversarial examples. Conventional attacks generate controlled noise-like perturbations that fail to reflect real-world scenarios and hard to interpretable. In contrast, recent unconstrained attacks mimic natural image transformations occurring in the real world for perceptible but inconspicuous attacks, yet compromise realism due to neglect of image post-processing and uncontrolled attack direction. In this paper, we propose RetouchUAA, an unconstrained attack that exploits a real-life perturbation: image retouching styles, highlighting its potential threat to DNNs. Compared to existing attacks, RetouchUAA offers several notable advantages. Firstly, RetouchUAA excels in generating interpretable and realistic perturbations through two key designs: the image retouching attack framework and the retouching style guidance module. The former custom-designed human-interpretability retouching framework for adversarial attack by linearizing images while modelling the local processing and retouching decision-making in human retouching behaviour, provides an explicit and reasonable pipeline for understanding the robustness of DNNs against retouching. The latter guides the adversarial image towards standard retouching styles, thereby ensuring its realism. Secondly, attributed to the design of the retouching decision regularization and the persistent attack strategy, RetouchUAA also exhibits outstanding attack capability and defense robustness, posing a heavy threat to DNNs. Experiments on ImageNet and Place365 reveal that RetouchUAA achieves nearly 100% white-box attack success against three DNNs, while achieving a better trade-off between image naturalness, transferability and defense robustness than baseline attacks.

No abstract

After the Boxer Rebellion ended with China’s crushing defeat and the signing of the Boxer Protocol, China participated in the 1904 Louisiana Purchase Exposition as its official debut at world’s fairs. The Chinese pavilion was supposed to represent the country’s national pride and cultural identity, yet ironically, the pavilion materialised the Chinese government’s weak position in its quasi-colonial relationship with the US – both politically and culturally – in terms of the appointment of architects, the design process, and the arrangement of construction. Such power interaction shaped an ambiguous ‘Chinese architecture’ presented at the fair, imitating the Beijing residence of a Chinese Prince while incorporating vernacular architectural elements from south China. It reflected the Chinese government’s early self-vision of its global image in an age of political turmoil and cultural uncertainty, and pioneered the exploration of an architectural ‘Chinese-ness’ in the early twentieth century.

Abstract. A 2D land administration system is insufficient for managing private properties and common property areas in a multi-story structure. Building information modelling (BIM) can be used to provide a clearer representation and more efficient management of the rights, restrictions, and responsibilities (RRR) inside buildings and address the challenges of 2D representations. However, a land surveyor should still draw the legal boundaries and group ownership spaces manually inside 3D BIM authoring tools. This research aims to provide an automatic approach to define three different types of legal boundaries and group the common properties and private properties within a building. This work contributes to the use and development of BIM by providing an automatic technique to creating property ownership, allowing for easier search and retrieval of 3D property information. More significantly, it can potentially minimize the time and cost of creating BIM-based 3D cadastral data for complex multi-story structures and improve the efficiency in urban land administration.