.png){kind=icon}
.png){kind=icon}
Return
Benchmarking Adversarial Patch Selection and Location
.png){kind=icon}
PRE
AI
DOI：10.3390/math14010103.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Adversarial patch attacks threaten the reliability of modern vision models. We present PatchMap, the first spatially exhaustive benchmark of patch placement, built by evaluating over 1.5x108 forward passes on ImageNet validation images. PatchMap reveals systematic hot-spots where small patches (as little as 2% of the image) induce confident misclassifications and large drops in model confidence. To demonstrate its utility, we propose a simple segmentation-guided placement heuristic that leverages off-the-shelf masks to identify vulnerable regions without any gradient queries. Across five architectures-including adversarially trained ResNet-50-our method boosts attack success rates by 8-13 percentage points compared to random or fixed placements.
Keywords:
adversarial patches
patch placement
spatial vulnerability map
ImageNet
robustness benchmarking
segmentation-guided placement
location-aware attacks
confidence drop
Journal
IF:
2.2
Papers:
2.9K
Citations:
3.6W