Abstract— This paper review deals with the job of creation of automatic on-screen translations for images, which is useful for many image related applications such as video and image retrieval along with the developing many tools that aid visually weakened or damaged persons to access picture-based information. Automatic image note methods are very helpful for organization systems, image search and retrieval. In this paper an analysis on automatic image caption generation using deep learning has been carried out. During the review, papers from 2015 to 2017 and one paper form 1996 has been thought about/believed for described/explained analysis specifically from journals only. The judging requirements describes the importance of an image and what composition properties must be satisfied may change/differ from user to user. More than that, the single judging requirements may compete against each other. The overall (related to the beautiful design and construction of buildings, etc.) of the model carefully studied remains complex, combining many ways of doing things and methods to deliver generating effective image description. The future scope may be working on text-to-speech technology, so that the created descriptions are automatically read out loud. Another focus maybe on translating videos directly to (series of words that make sense and that have a subject and a verb) instead of creating (written descriptions/on-screen translations) of images.
Keywords—Deep learning, automatic image caption generation, Artificial Intelligence, Information Storage and Retrieval, Image Note, Word-Based Indexing.
Automatic generation of textual descriptions for visual content has attracted extensive research interest in the fields of computer vision, multimedia and natural language processing. Image captioning, which aims at discovering objects and their relationship in a single image, as well as translating them with natural sentences, has been well studied. In the past it is witnessed an exponential growth in the amount of digital information available on the Internet. 17 Instagram, one of the best known photo sharing websites, hosts more than 16 billion images, for 130 million people. Browsing and finding pictures in large-scale and heterogeneous collections are an important problem that has attracted much interest within information retrieval.