A Novel Approach to Generate the Captions for Images with Deep Learning using CNN and LSTM Model
Sudesh Rao, Sethulakshmi Santhosh, Preethi Salian K, T Chidananda, Prathyakshini, S Sandeep Kumar · 2022
Everyday millions of images are circulated in internet, news, articles, documents, and many of those images are used for advertisement. End users are not interested in every image they come across. We also use internet to find the images but when we search, we will get lots of irrelevant images it will be difficult for us to categorise the images. E-commerce business is the one, which will analyse the images and generates the appropriate attributes for online catalogs. We could save time and cost if we can enable automatic product tagging. In this proposed model we can accomplish this job of image captioning. We do find CCTV everywhere today, It help us to see the world, but along with this if we are able to generate a caption for the images captured in CCTV, We can sense the dangerous activity and can raise the alarm. It will be very helpful if we have a way to automatically create captions for photos. Our proposed method will generate visual captions using the convolutional neural network (CNN) as well as the Long-Term Short-Term Memory (LSTM) model for In-depth Reading method.