A method for image-text matching based on semantic filtering and adaptive adjustment.