A scheme of long short-term memory (LSTM) -CNN features extraction and spatial relationship to anonymize image and video for Privacy-Preserving Human Activity Recognition of multiple sequences of data

• Main Authors: Kumar, Kambala Vijaya, Ayyappa, Yalanati, Vinay Kumar, B, Jonnadula, Harikiran, Tondepu, Yamini, Vidya, Nusumu Sree
• Format: Conference Proceeding
• Language: English
• ISSN: 2640-5768
• DOI: 10.1109/AISP57993.2023.10134958

Safeguarding security from stowed away pictures and recorded recordings is a significant social issue. To store information related picture or video data the arrangement of human movement can be recognized for examining the different issues of exercises. And yet guarantee that it doesn't record the significant video or catch the activities of human it is abusing the security of human. This article presents a crucial way to deal with tackling such clashing undertakings: acknowledgment of human movement while utilizing just mysterious low-goal recordings (Example: lSx16). We are presenting the Deep learning, CNN, Computer Vision OpenCV worldview, the calculation making sense of the insignificant picture transformation and to make a few low-goals of profoundly goal informative recordings from video. Our ideas is concentrating on different kinds of sub-pixel changes that are streamlined for arranging exercises, which permits the study hall expert to utilize existing high-goal recordings by bringing a few LR preparing recordings adjusted for this issue. The commonsense result results are affirmed that the OpenCV in PC models can profit from movement acknowledgment from very human action recordings. In this paper we are seeing as the negligible of human exercises from the pictures or recordings taken from the different public where the high goal can be distinguished and getting the exceptionally low goal and protection can be stowed away from such pictures or recordings.