Visual Analysis of Operators-Machinery Interaction

Industry 4.0 describe new paradigms for seamless interaction between humans and machines. The new industry relies on intelligent systems that are able to monitor the process flow of industrial production, the maintenance of the equipment and the quality of the manufacturing. In this scenario, Computer Vision and Deep Learning can be very important tools to improve industrial machinery, making their interaction with human operators easier and safer. From this idea comes the collaboration between UNIMORE/AImageLab and Tetra Pak for two research projects aimed at studying and prototyping new methods for the analysis of operator-machinery interactions. Tetra Pak UX experts and Engineers are researching innovative techniques for the human-centered design of new generation of machines, with the goal of improving operator performance, safety, well-being, and satisfaction. 

The first project comprises a novel set of deep learning and computer vision techniques to rethink the way in which human-machine interaction can be analyzed and evaluated. AImageLab and Tetra Pak are committed to jointly designing new models, collecting data and experimenting 3D pose estimation algorithms and posture analysis, based on the latest innovative AI techniques available in research. The second project focuses on improving the existing system, replacing the current re-identification module and enhancing the tracking, trajectory analysis and pose estimation modules, in order to handle multi-person scenarios. Action recognition techniques are also explored in order to enrich the analysis and the overall quality of the output.