ViraScience.
Scholarly record

Privacy-Preserving Egocentric Activity Recognition from Extreme Low Resolution

Michael S. Ryoo, Brandon Rothrock, Charles Fleming

Abstract

Privacy protection from video taken by wearable cameras is an important societal challenge. We desire a wearable vision system that can recognize human activities, yet not disclose the identity of the participants. Video anonymization is typically handled by decimating the image to a very low resolution. Activity recognition, however, generally requires resolution high enough that features such as faces are identifiable. In this paper, we propose a new approach to address such contradicting objectives: human activity recognition while only using extreme low-resolution (e.g., 16x12) anonymized videos. We introduce the paradigm of inverse super resolution (ISR), the concept of learning the optimal set of image transformations to generate multiple low-resolution videos from a single video. Our ISR learns different types of sub-pixel transformations optimized for the activity classification, allowing the classifier to best take advantage of existing high-resolution videos (e.g., YouTube videos) by generating multiple LR training videos tailored for the problem. We experimentally confirm that the paradigm of inverse super resolution is able to benefit activity recognition from extreme low-resolution videos (e.g., 16x12 and 32x24), particularly in first-person scenarios.

Similar items

  • A multi-task learning based hybrid prediction algorithm for privacy preserving human activity recognition framework

    Open-access article2021Bulletin of Electrical Engineering and Informatics['Vijaya Kumar Kambala', 'Harikiran Jonnadula', '']

    There is ever increasing need to use computer vision devices capture videos as part of many real-world applications. However, invading privacy people the cause concern. for protecting while are used purposefully based on objective functions. One such case human activity recogniti

  • Privacy preserving human activity recognition framework using an optimized prediction algorithm

    Open-access article2022IAES International Journal of Artificial Intelligence['Kambala Vijaya Kumar', 'J. Harikiran', '']

    Human activity recognition, in computer vision research, is the area of growing interest as it has plethora real-world applications. Inferring actions from one or more persons captured through a live video its immense utility contemporary era. Same time, protecting privacy humans

  • Presence detection and activity recognition using low-resolution passive IR sensors

    Scholarly record2013Oliver Amft, Luis Lopera, Frank Clermont

    Various installations and appliances used by building occupants during the day are manually operated, including office devices and kitchen appliances. Traditionally, activity monitoring systems require multi-modal sensor installations in order to monitor these types of rooms. The

  • Extreme Low Resolution Activity Recognition with Multi-Siamese Embedding Learning

    Scholarly record2017CoRRMichael S. Ryoo, Kiyoon Kim, Hyun Jong Yang

    This paper presents an approach for recognizing human activities from extreme low resolution (e.g., 16x12) videos. Extreme low resolution recognition is not only necessary for analyzing actions at a distance but also is crucial for enabling privacy-preserving recognition of human

  • Object-Centric Spatio-Temporal Pyramids for Egocentric Activity Recognition

    Scholarly record2013BMVCTomas McCandless, Kristen Grauman

    Activities in egocentric video are largely defined by the objects with which the camera wearer interacts, making representations that summarize the objects in view quite informative. Beyond simply recording how frequently each object occurs in a single histogram, spatio-temporal

  • VGAN-Based Image Representation Learning for Privacy-Preserving Facial Expression Recognition

    Scholarly record2018Jiawei Chen, Janusz Konrad, Prakash Ishwar

    Reliable facial expression recognition plays a critical role in human-machine interactions. However, most of the facial expression analysis methodologies proposed to date pay little or no attention to the protection of a user’s privacy. In this paper, we propose a Privacy-Preserv