Text is difficult to read in some cases due to text orientation, writing style, very light colors, etc. Visually impaired or visually impaired people have difficulty reading text in all of these situations. The architecture proposed in this work is intended to detect and identify Arabic characters with vowels in natural environments. This architecture can help visually impaired or blind people to read the text correctly. It allows users to read the text in a better and more immersive way by combining augmented reality with digital material. The approach uses both deep learning and more specifically the VGG 19 model and augmented reality to improve the efficiency, clarity, and accuracy of text reading. For text detection and identification we use the VGG 19 model, and for text visualization, we use augmented reality. The implementation technique presented in this research for an augmented reality interactive virtual assistant system is for users to use their smartphone's camera to receive enhanced text information via a text image and a three-dimensional image to understand the displayed text. It offers an interesting way to understand their environment. The use of augmented reality to better display recognized text in 3D is a fantastic feature. User research studies are conducted to assess usability and user satisfaction.