| Popis: |
Over the years, various attempts have been made to make social media and devices more disabled friendly. Screen readers and other assisting technologies have made it easier for people with visual disabilities like low vision and blindness to interact with social media. However, images on the social network are still inaccessible to them. Developing the description of an image in natural language is an upcoming problem which is at the intersection of disciplines of artificial intelligence, computer vision, and natural language processing. The image captioning task, also known as visual captioning, makes the technical chassis of many important applications of semantic visual search and visual questioning, photo and video sharing in social media, visual intelligence in chatting robots, and aid for visually impaired people to perceive surrounding objects and happenings. To describe an image effectively, it involves the detection of objects in images, identifying scenes and attributes of the objects. Then, these labels are used to construct semantically meaningful sentences to generate paragraphs which describe the images. Our research makes an attempt to carry out the task to produce a paragraph like a description of images. |
