neuralsec / daedalus-attack Goto Github PK

Hi,
i have tried your project and train your network on a specific image from COCO dataset.
eventually i got few files:
-best example of distortion
-Daedalus example batch .npz
-Distortion of image . npy
-X_adv. npy
-distortions.npy

how can i use them to see the detection before the attack and the final image after the attack?

thanks

Hi, thank you very much for the project,
would appreciate your help,
We're trying to run the adversarial examples from your site, as they are
over the original Yolo3 (the C version) and it does detect a person,
For example we tried taking 'Best\ example\ of\ 5\ Distortion\ 99.68110656738281.png' from this site, and run it as it is over the Original Yolo3, and we do see a person detection in Yolo's output.
Are we doing something wrong? or maybe the examples need to be run against the detector they were trained upon?

Best regards,
Blingo

Is there a requirements file for this project? I know its a few years old now.

Hi,
How can I download your paper?
I couldn't find it on the internet.

Hi, could you please provide the information about installation environment? I don't know which version of tensorflow you use.. And any other tips? Thanks!

Hi,

I would like to know if anyone has tested this on a Faster R-CNN framework?
I am currently doing my thesis and need to create adversarial attacks on satellite images using Faster-RCNN.

Hello,
I am attempting to run the l2_yolov3 section of code and coming across the error:
"AbortedError (see above for traceback): Operation received an exception:Status: 3, message: could not create a dilated convolution forward descriptor, in file tensorflow/core/kernels/mkl_conv_ops.cc:1111".
I have traced this error to some of the loss function initializations/calculations once X_adv, distortions = attacker.attack(X_test) is executed. It looks like the loss values are not being stored properly in a tensor. This issue might be tied with the newest version of TF& Keras, but any help regarding this issue would be greatly appreciated.
Thank you for your time.
-Lena