The Use of Deep Learning, Image Processing, and High-Performance Computing: A Systematic Mapping Study

Deep Learning (DL), combined with Image Processing (IP) and High-Performance Computing (HPC) provides a powerful tool for enhancing decision-making processes. This Systematic Mapping Study (SMS) explores the utilization of DL, IP, and HPC techniques in various fields. This SMS addresses four research questions (QR) related to the most used applications of these technologies, the most commonly used DL architectures for IP using HPC, and the ways in which DL architectures can be utilized for IP tasks. In this study, we analyzed a total of 134 papers published between 2010 and April 2023, focusing on the utilization of DL, IP, and HPC. The results revealed that these technologies were widely adopted in various fields including medicine, biology, computer science, geology, physical chemistry, industry, and others. The most common application was object detection, accounting for 18.96% of the cases studied, followed by recognition (16.37%) and classification (15.51%). Other applications included identification, analysis, segmentation, prediction, and diagnosis, as well as performance optimization and improvement. These findings demonstrate the prevalence of these deep learning architectures for image processing in high-performance computing environments. They also highlight the growing impact of these technologies in diverse fields, opening new avenues for research and application in these domains.