VCR-Bench: A Modular Open-Source Benchmark for Video Classification Robustness

查看原文 / View Original 编辑 / Edit

Computer Science > Computer Vision and Pattern Recognition arXiv:2610.08936 (cs) [Submitted on 6 Oct 2026] Title:VCR-Bench: A Modular Open-Source Benchmark for Video Classification Robustness Authors:Maksim Plinskiy, Aleksandr Gushchin, Sergey Lavrushkin, Dmitriy S. Vatolin, Anastasia Antsiferova

计算机科学 > 计算机视觉与模式识别 arXiv:2610.08936 (cs) [提交于 2026 年 10 月 6 日] 标题：VCR-Bench：一个用于视频分类鲁棒性的模块化开源基准测试作者：Maksim Plinskiy, Aleksandr Gushchin, Sergey Lavrushkin, Dmitriy S. Vatolin, Anastasia Antsiferova

Abstract: Robustness of image classification has several benchmarks, but their video counterparts are absent. In video classification temporal dimension introduces additional degrees of freedom for adversarial attacks, defenses, and preprocessing. Temporal sampling, perturbation budgets, and metric aggregation also interact in ways with no direct analogue in the image setting. Therefore, robustness for video classifiers is studied across scattered, incompatible implementations, making reported numbers hard to reproduce and analyze.

摘要：图像分类的鲁棒性已有多个基准测试，但视频领域尚无对应的基准。在视频分类中，时间维度为对抗攻击、防御和预处理引入了额外的自由度。时间采样、扰动预算和指标聚合的相互作用方式在图像设置中没有直接的对应物。因此，视频分类器的鲁棒性研究分散在互不兼容的实现中，导致报告的数据难以复现和分析。

We introduce VCR-Bench, a modular open-source benchmark framework that standardizes video loading, wrappers for classifiers, adversarial attacks and defenses, perceptual metrics, configuration presets, and result logging. VCR-Bench currently integrates 30 video classification models, 14 adversarial attacks, and 10 defense wrappers under a common evaluation protocol.

我们推出了 VCR-Bench，这是一个模块化的开源基准测试框架，它标准化了视频加载、分类器封装、对抗攻击与防御、感知指标、配置预设以及结果记录。VCR-Bench 目前在统一的评估协议下集成了 30 个视频分类模型、14 种对抗攻击和 10 个防御封装。

We evaluate representative video classifiers, attacks, and defenses on Kinetics-400 subset, reporting clean accuracy, attack success rate, perceptual quality, runtime, and memory usage. VCR-Bench is released with documented installation, reproducible run presets, component-extension interfaces, and scripts for reproducing the reported results at this https URL.

我们在 Kinetics-400 子集上评估了具有代表性的视频分类器、攻击和防御方法，并报告了干净准确率、攻击成功率、感知质量、运行时间和内存占用情况。VCR-Bench 发布时附带了详细的安装文档、可复现的运行预设、组件扩展接口以及用于复现报告结果的脚本，详见此 HTTPS 链接。