zxwu / adv_cloak

Code for Making an Invisibility Cloak: Real World Adversarial Attacks on Object Detectors

This is the code for Making an Invisibility Cloak: Real World Adversarial, which studies the transferability of adversarial attacks on object detectors.

It is built upon Maskrcnn-Benchmark. Please refer to Maskrcnn-Benchmark for installation instructions. The patch transformation code is modified upon Adversarial YOLO.

For training adversarial patches on a single model

See ./train.sh You need to download pretrained weights to begin with.

For training an ensemble of models.

tools/train_net_ensemble.py for training an ensemble of models.

For evaluating models

./eval_patch.sh