Social robots are increasingly becoming part of our society. In this context, the robots’ interaction capabilities depend on their ability to detect and respond to human cues and the stimuli in their surroundings as humans would. This paper presents a computer vision-based joint attention system using ROS that gathers user and environment information and computes the shared focus of attention dynamically in real-time during human-robot interaction. Our system clusters the detector’s data and includes an intensity-based management system that assigns and adjusts priority levels to each stimulus to compute the final focus of attention. By incorporating bio-inspired concepts, the system enables the robot to properly respond to non-verbal user cues such as gaze and hand gestures and to guide users’ attention towards the focus of interest while monitoring the achievement of joint attention. We have used ROS for the system implementation to provide our system with modularity to integrate each joined attention-related feature, simplify integration across robotic platforms, and extend the system’s capabilities in future iterations.