Image Captioning Using Deep Learning Model

Abstract

Image Captioning means that the natural language descriptions are generated automatically based on the content of an image. It's an important aspect of scene comprehension since it integrates computer vision and natural language processing knowledge. Numerous methods and algorithms are developed by researchers to increase the accuracy of image captioning. However, it is one of the major questions for future researchers to get optimized result in captioning an image. Furthermore, there are thousands of gray-scale images that are captioned. In this proposed work, different pre-trained models are used to extract features of images through Convolutional Neural Network (CNN) for colored images and gray-scale images from dataset and then, the extracted features are fed into LSTM, which generates caption for images. At last, the model’s accuracy of color and gray-scale images are studied to determine the model’s capability in captioning both types of images.