Single Run Action Detector over Video Stream -- A Privacy Preserving Approach

TL;DR

This paper introduces S-RAD, a real-time, privacy-preserving action detection system for video streams that achieves comparable accuracy to state-of-the-art methods while being computationally efficient enough for edge devices.

Contribution

The paper presents S-RAD, a novel end-to-end action detector combining Faster-RCNN with temporal modeling, optimized for privacy and real-time edge deployment.

Findings

Achieves comparable accuracy to state-of-the-art methods

Significantly lower model size and computational demand

Enables real-time processing on edge devices like Nvidia Jetson Xavier

Abstract

This paper takes initial strides at designing and evaluating a vision-based system for privacy ensured activity monitoring. The proposed technology utilizing Artificial Intelligence (AI)-empowered proactive systems offering continuous monitoring, behavioral analysis, and modeling of human activities. To this end, this paper presents Single Run Action Detector (S-RAD) which is a real-time privacy-preserving action detector that performs end-to-end action localization and classification. It is based on Faster-RCNN combined with temporal shift modeling and segment based sampling to capture the human actions. Results on UCF-Sports and UR Fall dataset present comparable accuracy to State-of-the-Art approaches with significantly lower model size and computation demand and the ability for real-time execution on edge embedded device (e.g. Nvidia Jetson Xavier).