ТЕНДЕНЦІЇ РОЗВИТКУ ГЕНЕРАЦІЇ ТЕКСТОВОГО ОПИСУ ЗОБРАЖЕНЬ

This paper covers advanced approaches and challenges in generating textual descriptions for images. Modern models, such as BLIP, CLIP, and others that use self-learning, contrastive learning, and noise robustness to improve the accuracy and reliability of descriptions are considered. The main trends and problems of the development of textual descriptions for images are viewed.