Akademik Veri Yönetim Sistemi
International Journal of Intelligent Systems and Applications in Engineering, cilt.11, sa.4, ss.816-830, 2023 (Scopus)
It is widely recognized that modern systems can discern the context of an image and enrich it with relevant captions through the fusion of computer vision and natural language processing, a technique referred to as 'image caption production.' This article aims to shed light on and analyze various image captioning techniques that have evolved over the past few decades, including the Attention Model, Region-Level Caption Detection, Semantic Content-Based Models, Multimodal Models, and more. The evaluation of these techniques employs diverse criteria such as Precision Rate, Recall Rate, F1 Score, Accuracy Rate, among others, while employing various datasets for comparison. This article offers a comprehensive structural examination of contemporary image captioning methods. Researchers can leverage the insights from this analysis to develop innovative, improved approaches that sidestep the shortcomings of older methods while retaining their beneficial aspects.