Abstract: To address the limited single-image information-processing method, insufficient global knowledge-processing capability of the traditional model, and the lack of explainability in detection ...