IAES International Journal of Artificial Intelligence (IJ-AI) Vol. 11, No. 1, March 2022, pp. 254~264 ISSN: 2252-8938, DOI: 10.11591/ijai.v11.i1.pp254-264 254 Journal homepage: http://ijai.iaescore.com Privacy preserving human activity recognition framework using an optimized prediction algorithm Kambala Vijaya Kumar, Jonnadula Harikiran School of Computer Science and Engineering, VIT-AP University, Amaravathi, India Article Info ABSTRACT Article history: Received Jul 23, 2021 Revised Dec 22, 2021 Accepted Jan 3, 2022 Human activity recognition, in computer vision research, is the area of growing interest as it has plethora of real-world applications. Inferring actions from one or more persons captured through a live video has its immense utility in the contemporary era. Same time, protecting privacy of humans is to be given paramount importance. Many researchers contributed towards this end leading to privacy preserving action recognition systems. However, having an optimized model that can withstand any adversary models that strives to disclose privacy information. To address this problem, we proposed an algorithm known optimized prediction algorithm for privacy preserving activity recognition (OPA-PPAR) based on deep neural networks. It anonymizes video content to have adaptive privacy model that defeats attacks from adversaries. The privacy model enhances the privacy of humans while permitting highly accurate approach towards action recognition. The algorithm is implemented to realize privacy preserving human activity recognition framework (PPHARF). The visual recognition of human actions is made using an underlying adversarial learning process where the anonymization is optimized to have an adaptive privacy model. A dataset named human metabolome database (HMDB51) is used for empirical study. Our experiments with using Python data science platform reveal that the OPA-PPAR outperforms existing methods. Keywords: Aadaptive privacy model Adversarial learning Deep neural networks Human action recognition Visual privacy This is an open access article under the CC BY-SA license. Corresponding Author: Vijaya Kumar Kambala School of Computer Science and Engineering, Vellore Institute of Technology, VIT-AP University Amaravathi, Vijayawada, Andhra Pradesh, India Email: kvkumar@pvpsiddhartha.ac.in 1. INTRODUCTION Video based surveillance has become an important computer vision application. It has plenty of applications in the real world. While video based surveillance in different domains is useful, it has potential risk in terms of privacy leakage. Therefore, many researchers contributed towards privacy preserving action recognition. Human action recognition is an important research area with rich set of methods with machine learning, deep learning and generative adversarial network (GAN) based models. Action recognition using deep learning, often supported by privacy preserving method, are explored in [1]–[6]. Lyu et al. [1] proposed a deep learning based method for privacy preserving framework with fair and decentralized approach. Rasim et al. [2] proposed a deep learning based model for privacy preserving approach to protect personal data. Weng et al. [3] proposed a deep learning model with blockchain for privacy protection. Lyu et al. [4] studied federated cloud models to achieve fair and privacy preserving approaches to solve problems. Kumar et al. [5] explored deep learning algorithms and resolution images besides spatial relationships to recognize human actions. Rajpur et al. [6] proposed a cloud-based service to achieve privacy preserving action recognition using deep convolution neural network (CNN) model.