Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Physical Adversarial Camouflage Through Gradient Calibration and Regularization
Authors: Jiawei Liang, Siyuan Liang, Jianjie Huang, Chenxi Si, Ming Zhang, Xiaochun Cao
IJCAI 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experimental results on various detection models, angles, and distances show that our method significantly surpasses the state-of-the-art, with an average attack success rate (ASR) increase of 13.46% across distances and 11.03% across angles. Furthermore, experiments in real-world settings confirm the method s threat potential, highlighting the urgent need for more robust autopilot systems less prone to spoofing. |
| Researcher Affiliation | Academia | 1School of Cyber Science and Technology, Sun Yat-sen University Shenzhen Campus, China 2Peng Cheng Laboratory, Shenzhen, China 3Nanyang Technological University, Singapore 4National Key Laboratory of Science and Technology on Information System Security, Beijing, China EMAIL, EMAIL, zm EMAIL, EMAIL |
| Pseudocode | Yes | Algorithm 1: Nearest Gradient Calibration (NGC) and Algorithm 2: Loss-Prioritized Gradient Decorrelation (LPGD) provide structured pseudocode blocks. |
| Open Source Code | No | The text does not explicitly state that the authors are releasing their code for the methodology described in this paper, nor does it provide a direct link to a code repository. |
| Open Datasets | Yes | Each model is pretrained on the COCO dataset and implemented using the MMDetection framework [Chen et al., 2019]. |
| Dataset Splits | No | Our training set comprises 20,000 images captured from diverse angles and distances to enhance texture generation. For evaluation, adversarial camouflage is applied to the vehicle within CARLA, with images captured at elevation angles of {0 , 5 , 10 , 15 , 20 , 30 , 45 , 60 }, along with 2-degree azimuth increments for a thorough 360 sweep. Extended evaluations encompass distances of {5, 7.5, 10, 12.5, 15} meters and five distinct weather conditions: noon, sunset, night, foggy, and rainy. This describes data generation for training and evaluation scenarios, but not explicit train/test/validation splits with percentages or counts from a fixed dataset. |
| Hardware Specification | Yes | We optimize the textures over three epochs, with all experiments performed on a single NVIDIA A100 80GB GPU. |
| Software Dependencies | No | We utilize the CARLA simulator [Dosovitskiy et al., 2017] to generate datasets. employing Modern GL [Dombi, 2020] for differentiable rendering and segmentation mask S generation. implemented using the MMDetection framework [Chen et al., 2019]. The paper mentions specific software tools (CARLA, Modern GL, MMDetection) and libraries, but does not provide specific version numbers for them. |
| Experiment Setup | Yes | Texture optimization is conducted using the Adam optimizer with a learning rate of 0.1, employing Modern GL [Dombi, 2020] for differentiable rendering and segmentation mask S generation. We optimize the textures over three epochs, with all experiments performed on a single NVIDIA A100 80GB GPU. |