With the rapid development of deep learning, Deep Neural Networks (DNNs) have been widely applied in various fields, including intelligent visual inspection in textile industrial manufacturing. However, current DNNs still face the challenge of adversarial examples (AEs). According to the information that the researcher can obtain, AEs can be categorized into three types: white-box, score-based black-box, and hard-label black-box. Among them, hard-label black-box AEs are recognized as the most meaningful and practical. Currently, most researches on AEs target image models, there are relatively few researches on video models. To close this gap, we improve the original Monte Carlo algorithm and innovatively propose a hard-label black-box adversarial example generation algorithm for video models, called VDA. Extensive experiments show that, compared to the algorithm based on the original Monte Carlo, VDA can improve success rate by nearly 6 times under the same conditions. This research provides a new perspective for evaluating the security of video-based monitoring systems in the textile industry.

adversarial examples, hard-label, black-box, video models, textile intelligent monitoring