Learning Coated Adversarial Camouflages for Object Detectors
Authors: Yexin Duan, Jialin Chen, Xingyu Zhou, Junhua Zou, Zhengyun He, Jin Zhang, Wu Zhang, Zhisong Pan
IJCAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments demonstrate the superiority of CAC over the existing attacks, and it shows impressive performance both in the virtual scene and the real world. |
| Researcher Affiliation | Collaboration | 1Department of Watercraft Power, Army Military Transportation University of PLA, Zhenjiang, China 2College of Command and Control Engineering, Army Engineering University of PLA, Nanjing, China 3The 28th Research Institute of China Electronics Technology Group Corporation, Nanjing, China 4Communication Engineering College, Army Engineering University of PLA, Nanjing, China 5Railway Transportation College, Hunan University of Technology, Zhuzhou, China |
| Pseudocode | Yes | Algorithm 1 Algorithm of CAC |
| Open Source Code | No | No explicit statement or link is provided for the open-sourcing of the authors' code for their methodology. The only link provided is for a demo video and a third-party YOLOv5 GitHub. |
| Open Datasets | Yes | The model is trained on the COCO2014 dataset [Lin et al., 2014]. and trained on the Pascal VOC2007 trainval set or the combined Pascal VOC-2007 and Pascal VOC-2012 trainval set [Everingham et al., 2015]. |
| Dataset Splits | No | The paper mentions using COCO2014 and Pascal VOC datasets, including 'Pascal VOC2007 trainval set', but does not explicitly state specific training/validation/test split percentages, sample counts, or detailed splitting methodology beyond referencing the standard dataset names. |
| Hardware Specification | No | The paper does not provide any specific hardware details such as GPU or CPU models, processor types, or memory amounts used for running experiments. |
| Software Dependencies | No | The paper mentions using Faster R-CNN, YOLOv3, and YOLOv5, but does not provide specific version numbers for any software dependencies or libraries. |
| Experiment Setup | No | The paper mentions setting the confidence score threshold to 0.3 and choosing 300 for the top-n number of proposals, but it does not provide concrete hyperparameter values such as learning rate, batch size, number of epochs, or optimizer details required to reproduce the training of the adversarial camouļ¬age. |