Automated Image Caption Generator for Visually Impaired Using VGG16 and LSTM

Living in an era of visual communication, the visually impaired often encounter barriers that limit their access to the vast array of visual content prevalent in everyday life. Image captioning is the process of creating suitable word sequence for any picture. Recent advances in neural networks have led to advancements in the implementation of picture captioning. In neural network architecture, convolutional neural networks, or CNNs, are used to create captions that explain the contents of images. In this paper, we propose that after training the VGG16 deep learning architecture using different photographs, the textual description of the image be converted to speech and transmitted to the user. After that, the generated captions of the image are converted into speech so that the visually impaired person can listen and understand the description of the picture, thus giving them an idea of their surroundings. The BLEU score metrics will be used to assess the quality and accuracy of the generated caption.