Abstract:

When we saw an image automatically our brain recognizes the content/description of an image. But how the system analyses or recognizes an image, is the problem that occurs in most of the scene understanding scenarios. Computer vision (CV) researchers focused on this problem and gave a solution to it by using Deep Learning algorithms. Deep Learning algorithms are the most powerful algorithms which help to make tasks accurately and efficiently without human intervention. In this paper, we are predicting the caption for an image through Deep Learning algorithms called CNN and RNN. A Convolutional Neural Network (CNN) is used to extract the features and objects from an image. These features are then fed into a particular RNN-based model known as a Long Short Term Memory (LSTM), which is in charge of correctly ordering the extracted features and providing an accurate description of the image in a language like English. Deep Learning systems have made picture captioning can be done easily with the help of a dataset. The dataset we used is the Flickr8k dataset which is developed by Shadab Hussain.

Description:

CNN, Deep Learning, Flickr8k, LSTM, RNN, Xception