| Popis: |
Describing the content of an image has been a fundamental problem of Machine learning that connects computer vision and natural language processing. In recent years, the task of object recognition has advanced at an exceptional rate which in turn has made image captioning that much better and easier. In this paper, we have discussed the usage of image captioning using deep learning for the visually impaired. We have used Convolutional Neural Networks along with Long Short-Term Memory to train and generate captions for images along with a text-to-speech engine which makes the experience of visually impaired users who are browsing the internet much smoother. We discuss how the model was implemented, its different components and modules along with a result analysis conducted on a set of outputs peer reviewed by our colleagues, friends and professors. |
