A Systematic Survey of Remote Sensing Image Captioning

Image captioning is a cross-disciplinary task to automatically generate textural descriptions for a given image using computer vision and natural language processing techniques. Remote sensing image captioning refers to the application of this task to remote sensing images taken from high altitude b...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autor principal: Beigeng Zhao
Formato: article
Lenguaje:EN
Publicado: IEEE 2021
Materias:
Acceso en línea:https://doaj.org/article/3f8fd78812f74ad6ba3939b187eb4788
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
Descripción
Sumario:Image captioning is a cross-disciplinary task to automatically generate textural descriptions for a given image using computer vision and natural language processing techniques. Remote sensing image captioning refers to the application of this task to remote sensing images taken from high altitude by satellites, aircraft or drones. This interesting and valuable topic has only emerged in recent years and attracted considerable research attention. There has been extensive related work in the literature, with considerable results and an independent body of research, and various issues must be addressed in future work. However, to the best of our knowledge, there has been no review study in this area that can provide researchers with systematic reference information, which is the motivation of this study. To achieve this goal, 30 relevant articles were conditionally filtered and obtained for the review study. We analyzed and summarized the existing work from various perspectives, including technical solutions, data, evaluation metrics, and the experimental results of state-of-the-art methods. Based on this summary, the trends, pros and cons of the existing studies, issues to be addressed and valuable research directions in future work are discussed. The results of this paper can provide valuable reference information for researchers in related fields.