Natural Language Processing of Text-Based Metrics for Image Captioning

A rapidly increasing interest has been focused on natural language processing and computer vision study by automatically producing descriptive sentences for images in present trends. Image processing is a success factor, especially involving somaticized image knowledge and creating reliable, appropriately organized explanation phrases. In this paper, a dynamic program that uses the VCG16 platform to develop image descriptions and a long short-term memory (LSTM) to contain appropriate words with generated text keywords successfully is proposed. This paper determines the NLP method’s usefulness using Flickr8 K-statistical models and demonstrates that their model delivers accurate performance compared to the Bleu metric. The Bleu metric is an automated system for measuring a machine translation’s performance by monitoring the effectiveness of text transformed from natural to other languages. The results have been justified with images that satisfy the NLP algorithm.