Authors: Naufal Suryanto, Yongsu Kim, Harashta Tatimma Larasati, Hyoeun Kang, Thi-Thu-Huong Le, Yoonyoung HONG, Howon Kim

Conference: ICCV 2023

Abstract: Adversarial camouflage methods have garnered increasing attention due to their ability to attack object detectors from any viewpoint by covering the entire object's surface. In this paper, we present Adversarial Camouflage for Transferable and Intensive Vehicle Evasion (ACTIVE), a state-of-the-art physical camouflage attack framework designed to generate universal and robust adversarial camouflage capable of concealing any 3D vehicle from detectors. Our framework incorporates several innovative techniques to enhance universality and robustness. First, we employ a sophisticated texture rendering method that enables the application of common textures to different vehicles, without being constrained to a specific texture map. Subsequently, we introduce a novel stealth loss function that renders the vehicle undetectable, and utilize a smooth and a camouflage loss that enhance the naturalness of the adversarial camouflage. Our extensive experiments on 15 different models show that ACTIVE consistently outperforms other adversarial camouflage methods on various publicly available detectors, including the latest YOLOv7. Notably, our universality evaluations reveal promising transferability to other vehicle classes, tasks (i.e., segmentation models), and to the real world, not just to other vehicles.

Link: