Deep learning for overlapping objects detection with noise: a bibliometric analysis

Abstract

This study conducts a bibliometric analysis to investigate the utilization of deep learning for detecting overlapping objects in noisy environments. Despite the advancements in deep learning, accurately detecting overlapping objects amidst noise remains a significant challenge. Relevant publications were identified using the Scopus database, and Scopus and VOSviewer were utilized for analysis. The results reveal a notable increase in research interest in this area within these past few years, highlighting the significance of the topic. Key themes identified include novel network architectures, data augmentation techniques, and evaluation metrics. However, challenges such as robustness to noise and real-time performance persist. Collaborations among researchers and institutions are observed, emphasizing the interdisciplinary nature of the research field. The results of this study provide important insights for future research areas and solutions in recognizing overlapping objects with noise. Addressing the problems discussed above necessitates continued multidisciplinary research and the creation of more robust and productive deep learning systems. By leveraging the insights from this study, researchers can contribute to advancements in overlapping objects detection, leading to improved performance and applicability in various domains, including medical imaging, surveillance systems, and autonomous driving.