Recognition of Image Caption Generation Using Deep Neural Framework

Abstract

Computer vision and natural language processing techniques are required to generate textual descriptions of a given image, a task known as image captioning. Over the past ten years, research on Deep Learning and Neural Networks has upsurge as a result of the improved results. Nowadays, image processing is the main way to gather information from images, process them for a purpose, and performing operations on them. It also helps in finding a lot of information from a single image. This paper presents Recognition of Image Caption Generation Using Deep Neural Framework. By Flickr 8k dataset is used to experimentally evaluate the proposed methods. The goal of this paper is to use deep learning to detect, produce, and recognise meaningful captions for a given image. The Principal Component Analysis (PCA) method is used to extract the image's features. In this paper, RNN is used to detect, identify images, and generate captions from them. This model is evaluated based on standard evaluation metrics such as performance Time, Loss, prediction Accuracy, and BLEU (Bilingual Evaluation Understudy). Experimental results show that the described model achieves impressive performance compared to strong baseline methods.