Composed Image Retrieval

Composed Image Retrieval (CoIR) is a task in the field of computer vision that aims to retrieve images from a large database based on queries composed of multiple elements such as text, images, and sketches. The goal of this task is to develop algorithms capable of understanding and integrating information from multiple sources to accurately match the query images, thereby enhancing user expressiveness. CoIR has significant application value in scenarios like content creation, image search, and personalized recommendations.