MVTamperBench: Evaluating Robustness of Vision-Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Agarwal, Amit, Panda, Srikant, Charles, Angeline, Kumar, Bhargava, Patel, Hitesh, Pattnayak, Priyaranjan, Rafi, Taki Hasan, Kumar, Tejaswini, Meghwani, Hansa, Gupta, Karan, Chae, Dong-Kyu
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909645508968448
author Agarwal, Amit
Panda, Srikant
Charles, Angeline
Kumar, Bhargava
Patel, Hitesh
Pattnayak, Priyaranjan
Rafi, Taki Hasan
Kumar, Tejaswini
Meghwani, Hansa
Gupta, Karan
Chae, Dong-Kyu
author_facet Agarwal, Amit
Panda, Srikant
Charles, Angeline
Kumar, Bhargava
Patel, Hitesh
Pattnayak, Priyaranjan
Rafi, Taki Hasan
Kumar, Tejaswini
Meghwani, Hansa
Gupta, Karan
Chae, Dong-Kyu
contents Multimodal Large Language Models (MLLMs), are recent advancement of Vision-Language Models (VLMs) that have driven major advances in video understanding. However, their vulnerability to adversarial tampering and manipulations remains underexplored. To address this gap, we introduce \textbf{MVTamperBench}, a benchmark that systematically evaluates MLLM robustness against five prevalent tampering techniques: rotation, masking, substitution, repetition, and dropping; based on real-world visual tampering scenarios such as surveillance interference, social media content edits, and misinformation injection. MVTamperBench comprises ~3.4K original videos, expanded into over ~17K tampered clips covering 19 distinct video manipulation tasks. This benchmark challenges models to detect manipulations in spatial and temporal coherence. We evaluate 45 recent MLLMs from 15+ model families. We reveal substantial variability in resilience across tampering types and show that larger parameter counts do not necessarily guarantee robustness. MVTamperBench sets a new benchmark for developing tamper-resilient MLLM in safety-critical applications, including detecting clickbait, preventing harmful content distribution, and enforcing policies on media platforms. We release all code, data, and benchmark to foster open research in trustworthy video understanding. Code: https://amitbcp.github.io/MVTamperBench/ Data: https://huggingface.co/datasets/Srikant86/MVTamperBench
format Preprint
id arxiv_https___arxiv_org_abs_2412_19794
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle MVTamperBench: Evaluating Robustness of Vision-Language Models
Agarwal, Amit
Panda, Srikant
Charles, Angeline
Kumar, Bhargava
Patel, Hitesh
Pattnayak, Priyaranjan
Rafi, Taki Hasan
Kumar, Tejaswini
Meghwani, Hansa
Gupta, Karan
Chae, Dong-Kyu
Computer Vision and Pattern Recognition
68T37, 68T05, 68Q32, 68T45, 94A08, 68T40, 68Q85
I.2.10; I.2.7; I.5.4; I.4.9; I.4.8; H.5.1
Multimodal Large Language Models (MLLMs), are recent advancement of Vision-Language Models (VLMs) that have driven major advances in video understanding. However, their vulnerability to adversarial tampering and manipulations remains underexplored. To address this gap, we introduce \textbf{MVTamperBench}, a benchmark that systematically evaluates MLLM robustness against five prevalent tampering techniques: rotation, masking, substitution, repetition, and dropping; based on real-world visual tampering scenarios such as surveillance interference, social media content edits, and misinformation injection. MVTamperBench comprises ~3.4K original videos, expanded into over ~17K tampered clips covering 19 distinct video manipulation tasks. This benchmark challenges models to detect manipulations in spatial and temporal coherence. We evaluate 45 recent MLLMs from 15+ model families. We reveal substantial variability in resilience across tampering types and show that larger parameter counts do not necessarily guarantee robustness. MVTamperBench sets a new benchmark for developing tamper-resilient MLLM in safety-critical applications, including detecting clickbait, preventing harmful content distribution, and enforcing policies on media platforms. We release all code, data, and benchmark to foster open research in trustworthy video understanding. Code: https://amitbcp.github.io/MVTamperBench/ Data: https://huggingface.co/datasets/Srikant86/MVTamperBench
title MVTamperBench: Evaluating Robustness of Vision-Language Models
topic Computer Vision and Pattern Recognition
68T37, 68T05, 68Q32, 68T45, 94A08, 68T40, 68Q85
I.2.10; I.2.7; I.5.4; I.4.9; I.4.8; H.5.1
url https://arxiv.org/abs/2412.19794