Picture Tales: An Approach for Story Generation Using a Series of Images

Visual storytelling is a challenging problem in artificial intelligence that requires machines to understand a sequence of images to produce a narrative story. It takes image captioning a step forward and is a more complex task as it deals with coherence, continuity, and creativity in the generated text to form a complete story. In this paper, we put forth an approach based on the encoder-decoder architecture to generate a story for a sequence of images. We take into consideration the images and their respective descriptions to create coherent stories. We evaluate the performance of our model on the VIST dataset [1] and achieve highly comparable results with respect to the existing systems.