A Deep Learning for the Generation of Textual Story Corresponding to a Sequence of Images
International Journal of Recent Technology and Engineering (IJRTE) · 2019
Generating a short story for a sequence of images is much more interesting than generating a single line textual description for an image. Story generation involves relating the meaning of the previous image and the current image and continuing this through out the sequence of images. This can be helpful for better understanding of the situation. In this paper we present our idea of generating story using a CNN model which is pre trained on MSCOCO dataset that can detect objects and concepts of language modelling and NLP text pre-processing techniques . We used a custom stories dataset in which we manually labelled every sentence in every story. Number of sentences in the generated story is equal to the number of images. The results are quite accurate in many cases for a small custom stories dataset and the performance is expected to increase with a bigger dataset.