Visual assistance camera (VAC)

Abstract

The advancement made in machine learning has allowed us to use neural networks in practical scenarios. We use convolutional neural networks and long short term memory units to make out the features in the image and then generate a caption of thatimage. The product is a project that uses deep neural networks to firstly give dense image descriptions and then give its output to the visually impaired user in audio format. This product will facilitate the visually impaired person so that he can have better sense of what is in front of him. Moreover, all the functionalities of the model will be incorporated in a web based mobile application that user will be able to use readily and easily. We use image data along with their annotations that is used to train the model on the dataset, the accuracy for the model depends on the range of images we have as well as the number of them and a combination of other factors.