Abstract: Visual Question Answering (VQA) systems witnessed a significant advance in recent years due to the development of large-scale Vision-Language Pre-trained Models (VLPMs). As the application ...