For object detection models, ([DetrForObjectDetection]), the model expects a list of dictionaries with a
class_labels and boxes key where each value of the batch corresponds to the expected label and number of bounding boxes of each individual image. | 
