Deep Learning-Based Tracking System for the Transition from Unhealthy to Healthy Lifestyles based on Indications Found in Social Media

Authors

  • Digvijay Singh, Sachin Sharma, Rahul Bhatt

DOI:

https://doi.org/10.17762/msea.v71i4.687

Abstract

Both the data compression and the data analysis research disciplines have been given a boost as a direct result of the widespread use of high-definition CCTV camera video. A third problem for the community that deals with video surveillance is the rising awareness of individuals about the sensitivity of their private information. This difficulty generates a need for the community to additionally include privacy protection. In this research, we propose a deep learning-based object tracking method that makes use of compressed domain residual frames in an effort to address the aforementioned demands. The aim is to be able to provide a public visual representation for data analysis that is respectful of individuals' right to privacy. In this piece of work, we investigate a situation in which the tracking is done directly on a constrained portion of the information recovered from the compressed domain. For the purposes of training and testing our network, we make sole use of the residual frames that have already been produced by the video compression codec. This very condensed representation also functions as an information filter, reducing the amount of confidential information that can be gleaned from watching a video stream. We are able to demonstrate that the use of residual frames for object tracking based on deep learning can be just as successful as the use of traditionally decoded frames in the same situations. To be more specific, the use of residual frames is especially useful in straightforward video surveillance situations that include continuous flow that does not overlap.

Downloads

Published

2022-09-02

How to Cite

Digvijay Singh, Sachin Sharma, Rahul Bhatt. (2022). Deep Learning-Based Tracking System for the Transition from Unhealthy to Healthy Lifestyles based on Indications Found in Social Media. Mathematical Statistician and Engineering Applications, 71(4), 1625–1633. https://doi.org/10.17762/msea.v71i4.687

Issue

Section

Articles