Privacy Protection for Social Video via Background Estimation and CRF-Based Videographer's Intention Modeling

Abstract

SUMMARYThe recent popularization of social network services (SNSs), such as YouTube, Dailymotion, and Facebook, enables people to easily publish their personal videos taken with mobile cameras. However, at the same time, such popularity has raised a new problem: video privacy. In such social videos, the privacy of people, i.e., their appearances, must be protected, but naively obscuring all people might spoil the video content. To address this problem, we focus on videographers' capture intentions. In a social video, some persons are usually essential for the video content. They are intentionally captured by the videographers, called intentionally captured persons (ICPs), and the others are accidentally framed-in (nonICPs). Videos containing the appearances of the non-ICPs might violate their privacy. In this paper, we developed a system called BEPS, which adopts a novel conditional random field (CRF)-based method for ICP detection, as well as a novel approach to obscure non-ICPs and preserve ICPs using background estimation. BEPS reduces the burden of manually obscuring the appearances of the non-ICPs before uploading the video to SNSs. Compared with conventional systems, the following are the main advantages of BEPS: (i) it maintains the video content, and (ii) it is immune to the failure of person detection; false positives in person detection do not violate privacy. Our experimental results successfully validated these two advantages. key words: intentionally captured person, conditional random field, background estimation, privacy protection, social video

Finding Important People in a Video Using Deep Neural Networks with Conditional Random Fields

“…ACC (%) REC (%) PRE (%) FPR (%) F1 (%) pared our approach to raw support vector machine decisions (SVM) and the previous work in [20] (SVM-CRF), which uses a support vector machine to obtain decision values and applies CRF to them together with features. Since the results in [20] shows that the improvement by the temporal consistency term in their model is not very large, we employed an SVM-CRF model simplified by removing the temporal consistency term. We tuned the hyperparameters of our DNNbased models and SVM-based models (i.e., learning rate, dropout ratio, weight decay ratio, and unit size of hidden layer N for DNN-based models, and γ and C of SVM with the radial basis function) with Bayesian optimization.…”

“…Recent progress in large scale datasets [11], [12] and DNN techniques have significantly improved the performance of various vision tasks, such as object classification [13]- [15] and semantic segmentation [16]- [19]. In this work, we also develop a deep model to classify people into important or unimportant ones, which is an extension of [5], [20]. As in these work, we uses a CRF built upon a deep model.…”