Automatic Image Captioning Using Neural Network

Abstract

    The Automatic Image Captioning aims towards generating natural and simple language sentences to describe the content of an Image. Image Captioning has gained a lot of interest in recent times. Generating accurate captions is a very challenging task. Although various solutions have been generated using the traditional CNN and RNN neural network algorithm for Image Captioning, still a problem arises is that the model can only generate Captions for the Image Concepts, which are seen in the Training Data set. In this paper, we propose a 2 Stage Model Architecture methodology with Deep Convolutional Network, which is known as the CNN algorithm for extracting the features and Long Short Term Memory (LSTM) algorithm to generate accurate Image Caption and to overcome the problems arisen in the traditional neural algorithms methodology. We evaluate our approach on the Flicker8k data set.