"VocPix: Image Captioning using CNN and LSTM"

Computer Engineering

With billions of users on sites like Facebook, Twitter, Instagram, and YouTube, social media has become an essential component of modern life. Social media has fundamentally altered how we communicate, share information, and engage with one another. It has also changed how we both consume and produce material. Therefore, an image caption generator has become essential in today's culture because it is necessary for social media addicts or people who are blind. It is a kind of algorithm or software program that makes use of Deep Learning techniques and analyses an image's visual information before converting it into plain language. It can be used as a plugin on the popular social networking sites of today to suggest appropriate captions for users to include with their postings. The goal of the suggested research is to create an image caption, also known as a description of an image, and to translate it into different languages, using CNN-LSTM architecture. To use the current word as input for the prediction of the following word, CNN layers will help in retrieving input data, and LSTM will extract essential information as it processes input. Python 3 and machine learning will be the programming languages used. This study will go into great detail about the many Neural networks involved, including their structures and functions. The program that converts created captions into spoken words is called a text-to-speech synthesizer, and it uses Natural Language Processing (NLP) to analyze and process the text. The text is subsequently converted into a synthesized speech representation using digital signal processing (DSP) technology. Here, we've created a practical text-to-speech synthesizer in the form of an easy-to-use application that reads aloud generated captions as synthesized speech. The proposed deep learning approach aims at generating the best caption for a particular image by analyzing and extracting various features from images and converting that textual caption into speech using Text-To-Speech (TTS).

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.