Story Visualization: Generation of Scenes sequentially for a given fable using NLP and Image Processing

Humans have a usual tendency to paint pictures in their minds when a story is being told. Undoubtedly, a picture conveys information more effectively than words. So it is necessary to instill visualization capability at an early age. In this project Story Visualization, we have aimed to improve the visualization of fables along with narration of the story. This is achieved by performing pronoun replacement to enhance story character extraction and improving fragmentation of the story to accurately depict the story, which is received as input in the form of text. The extracted attributes for scene generation are then sent to the image processing model which chooses appropriate images from the labeled dataset, transforms them relatively, and positions them based on grids, thereby generating a meaningful scene that nearly depicts the origin segment. Furthermore, the story is narrated, and the interactions between the story characters are uniquely voiced while they are exhibited through speech balloons. The comprehensive website presented abstracts NLP model, image processing model, and TextToSpeech model while giving a fairly simple user interface for users to interact with. To cover a broader audience, we have experimented in Indian languages and as proof of concept, Kannada language fables are visualized.