This application generates a caption of an image based on detected objects in the image. If the number of objects is not less than two, the application generates the caption applying the T5 model, if less, the application uses the VitGPT model.