image-retrieval-robustness

Experiments with the CLIP and LLaVA model using adversarial patch attacks.

Notebooks include the main algorithm of obtaining adversarial patches (in main.ipynb file) and models inference setup for additional experiments

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
LLaVA_OneVision_video_attacks_ipynb_.ipynb		LLaVA_OneVision_video_attacks_ipynb_.ipynb
README.md		README.md
for_video_fix_names.ipynb		for_video_fix_names.ipynb
llava-ppy.ipynb		llava-ppy.ipynb
main.ipynb		main.ipynb
one_vision_llava_hf.ipynb		one_vision_llava_hf.ipynb
patch_att_on_video.ipynb		patch_att_on_video.ipynb
patch_video.ipynb		patch_video.ipynb

Provide feedback