Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/VITA-Group/3D_Adversarial_Logo
[Preprint] "Can 3D Adversarial Logos Cloak Humans?"
https://github.com/VITA-Group/3D_Adversarial_Logo
3d-mesh adversarial-attacks cloak detection detector logos
Last synced: 12 days ago
JSON representation
[Preprint] "Can 3D Adversarial Logos Cloak Humans?"
- Host: GitHub
- URL: https://github.com/VITA-Group/3D_Adversarial_Logo
- Owner: VITA-Group
- License: mit
- Created: 2020-06-21T17:50:06.000Z (about 4 years ago)
- Default Branch: master
- Last Pushed: 2021-12-30T12:41:46.000Z (over 2 years ago)
- Last Synced: 2024-02-29T10:34:06.004Z (4 months ago)
- Topics: 3d-mesh, adversarial-attacks, cloak, detection, detector, logos
- Language: Python
- Homepage:
- Size: 3.33 MB
- Stars: 17
- Watchers: 7
- Forks: 3
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Lists
- awesome-yolo-object-detection - VITA-Group/3D_Adversarial_Logo - Group/3D_Adversarial_Logo?style=social"/> : 3D adversarial logo attack on different3D object meshes to fool a YOLOV2 detector. "Can 3D Adversarial Logos Clock Humans?". (**[arXiv 2020](https://arxiv.org/abs/2006.14655)**) (Applications)
README
# Can 3D Adversarial Logos Clock Humans? #
[](https://opensource.org/licenses/MIT)
[Can 3D Adversarial Logos Clock Humans?]()
Yi Wang\*, Jingyang Zhou*, Tianlong Chen, Sijia Liu, Shiyu Chang, Chandrajit Bajaj, Zhangyang Wang
## Overview
Examples of our 3D adversarial logo attack on different3D object meshes to fool a YOLOV2 detector.
## Methodology
## Prerequisites and Installation
- Python 3
- Pytorch 1.1.0
- CUDA 9.2 (lower versions may work but were not tested)
- TensorboardX
```shell
pip install tensorboardX tensorboard
```
- Neural renderer
```shell
pip install neural-renderer-pytorch
```
- Clone this repo
```shell
git clone https://github.com/TAMU-VITA/3D_Adversarial_Logo.git
cd 3D_Adversarial_Logo
```
- Yolov2 weights
```shell
mkdir weights
curl https://pjreddie.com/media/files/yolov2.weights -o weights/yolo.weights
```
## Usage
### Prepare the dataset
- Datasets and our prediction images can be found [here](https://drive.google.com/file/d/1CUC_Yy10sjGSvSG8wt-ToPtOJ_eEf8x-/view?usp=sharing).
- Our pre-trained 3d adversarial logos can be found [here](https://drive.google.com/file/d/1AeWOzhLUbf7XK4-bo-rtvcQ1VKO_bn30/view?usp=sharing).
Remark. After downloading them, put the data and weight under the `3D_Adversarial_Logo` folder.
### Arguments
- `--width` : the width of the universal logo.
- `--height` : the height of the universal logo.
- `--depth` : the depth of the universal logo (default is 1 to form an image).
- `--angle_range` : the angle range of the camera view for training (eg. 10 for +/- 10 degree)
- `--logonum` : the total faces of the 3D logo.
- `--train_mesh` : which mesh to train (the 123 denotes mesh 1&2&3, which can be further upgraded by adding `nargs='+'` in parsing args).
- `--test_mesh` : which mesh to test (the 123 denotes mesh 1&2&3, which can be further upgraded by adding `nargs='+'` in parsing args).
- `--consistent` : whether to share the same universal logo among all the meshes.
- `--logo_ref` : which logo to train with.
- `--train_patch` : whether to train 2D patch.
- `--restore_model` : whether to restore the saved universal logo.
- `--save_model` : whether to save the universal logo.
### Run 3D adversarial logos ###
- logo: X
```shell
python train_patch1.py --width=100 --height=150 --depth=1 --angle_range=0 --logonum=25 --train_mesh=123 --consistent --logo_ref=X --save_model >logo-records/logoX.out 2>&1 &
```
- logo: X (medium)
```shell
python train_patch1.py --width=100 --height=150 --depth=1 --angle_range=0 --logonum=10 --train_mesh=123 --consistent --logo_ref=X --save_model >logo-records/logoX.out 2>&1 &
```
- logo: X (small)
```shell
python train_patch1.py --width=100 --height=150 --depth=1 --angle_range=0 --logonum=5 --train_mesh=123 --consistent --logo_ref=X --save_model >logo-records/logoX.out 2>&1 &
```
- logo: H
```shell
python train_patch1.py --width=100 --height=150 --depth=1 --angle_range=0 --logonum=25 --train_mesh=123 --consistent --logo_ref=H --save_model >logo-records/logoH.out 2>&1 &
```
- logo: T
```shell
python train_patch1.py --width=100 --height=150 --depth=1 --angle_range=0 --logonum=25 --train_mesh=123 --consistent --logo_ref=T --save_model >logo-records/logoT.out 2>&1 &
```
- logo: G
```shell
python train_patch1.py --width=100 --height=100 --depth=1 --angle_range=0 --logonum=17 --train_mesh=123 --consistent --logo_ref=G --save_model >logo-records/logoG.out 2>&1 &
```
- logo: C
```shell
python train_patch1.py --width=100 --height=100 --depth=1 --angle_range=0 --logonum=17 --train_mesh=123 --consistent --logo_ref=C --save_model >logo-records/logoC.out 2>&1 &
```
- logo: O
```shell
python train_patch1.py --width=100 --height=100 --depth=1 --angle_range=0 --logonum=17 --train_mesh=123 --consistent --logo_ref=O --save_model >logo-records/logoO.out 2>&1 &
```
- logo: tw
```shell
python train_patch1.py --width=100 --height=100 --depth=1 --angle_range=0 --logonum=25 --train_mesh=123 --consistent --logo_ref=tw --save_model >logo-records/logotw.out 2>&1 &
```
- logo: drop
```shell
python train_patch1.py --width=100 --height=150 --depth=1 --angle_range=0 --logonum=25 --train_mesh=123 --consistent --logo_ref=drop --save_model >logo-records/logodrop.out 2>&1 &
```
### Train 2D adversarial patch
- logo: G
```shell
python train_patch1.py --width=100 --height=100 --depth=1 --angle_range=0 --logonum=17 --train_mesh=123 --consistent --logo_ref=G --restore_model --save_model --train_patch
```
## Citation
If you are use this code for you research, please cite our paper.
```
@article{chen2020can,
title={Can 3D Adversarial Logos Cloak Humans?},
author={Chen, Tianlong and Wang, Yi and Zhou, Jingyang and Liu, Sijia and Chang, Shiyu and Bajaj, Chandrajit and Wang, Zhangyang},
journal={arXiv preprint arXiv:2006.14655},
year={2020}
}
```