AcTrak: Controlling a Steerable Surveillance Camera using Reinforcement Learning

Abstract

Steerable cameras that can be controlled via a network, to retrieve telemetries of interest have become popular. In this paper, we develop a framework called AcTrak , to automate a camera’s motion to appropriately switch between (a) zoom ins on existing targets in a scene to track their activities, and, (b) zoom out to search for new targets arriving to the area of interest. Specifically, we seek to achieve a good trade-off between the two tasks, i.e., we want to ensure that new targets are observed by the camera before they leave the scene, while also zooming in on existing targets frequently enough to monitor their activities. There exist prior control algorithms for steering cameras to optimize certain objectives; however, to the best of our knowledge, none have considered this problem, and do not perform well when target activity tracking is required. AcTrak automatically controls the camera’s PTZ configurations using reinforcement learning (RL), to select the best camera position given the current state. Via simulations using real datasets, we show that AcTrak detects newly arriving targets 30% faster than a non-adaptive baseline and rarely misses targets, unlike the baseline which can miss up to 5% of the targets. We also implement AcTrak to control a real camera and demonstrate that in comparison with the baseline, it acquires about 2 × more high resolution images of targets.

Towards Fully Autonomous Drone Tracking by a Reinforcement Learning Agent Controlling a Pan–Tilt–Zoom Camera

“…We have identified a gap in the literature-to the best of our knowledge there is no existing literature on tracking drones using an RL agent controlling a PTZ camera. While similar approaches of using RL for PTZ control have recently been attempted for the applications of tracking cars [16] or humans [17], we believe that the task of drone tracking presents unique challenges, and to our knowledge has not been attempted.…”

“…Some elements of this solution might be AI-based (e.g., the detection part). • Automated (RL) PTZ camera control [2,16,17]: An end-to-end AI system that takes the camera image as an input and outputs the control request to the PTZ camera trained using RL.…”

“…They successfully transfer their model to real-life scenarios and run their model on-board of a Raspberry Pi. Fahim et al [17] also present an RL model for controlling a PTZ for the purpose of tracking human subjects. They train their policy within a simulator and evaluate it in real-world scenarios.…”

Active Visual Perception Enhancement Method Based on Deep Reinforcement Learning

“…Kim et al [20] leveraged object position and size information from video analysis systems to automatically control PTZ cameras, enhancing the accuracy of identifying abnormal behavior in video surveillance systems. Fahim et al [21] utilized an object detection model to detect targets in surveillance video frames, transforming them into a reinforcement learning state. By adaptively adjusting the camera's position and scaling level, they achieved tracking of existing targets and searching for new ones.…”