Conversational AI for Blind Users: Image Recognition and Assistance Using LSTM

Abstract

Creating evocative text summaries for photographs is the challenge of image captioning, a sophisticated computer vision task. This technology understands the content of an image and conveys it in a way that is legible by humans by combining concepts from computer vision and natural language processing. Blind people's need for image captioning stems from the basic need for equality of access to information and inclusivity. In order to enable people with visual impairments to understand visual content that would otherwise be unavailable to them, this technology is essential in giving descriptive textual information about the contents of images. Image captioning promotes a sense of autonomy and lessens reliance on sighted help by enabling blind people to autonomously explore and comprehend the visual elements of their surroundings. The requirements of the visually handicapped are being met by a number of current programs and systems that use image captioning technology. In order to improve accessibility for those with visual impairments, this research proposes a novel method that uses Convolutional Neural Network (CNN) techniques to construct an image captioning system. In order to help the blind understand visual content they frequently come across in their daily lives, the system attempts to provide meaningful and thorough descriptions of visuals. By utilizing CNNs, the model is able to extract and interpret pertinent characteristics from images, producing meaningful captions that are subsequently conveyed to users via assistive technologies like speech synthesis. By providing a possible solution to close the visual information gap and enable people with visual impairments to interact and navigate the visual world more effectively, the project tackles the critical need for inclusive technology.