Efficient Multi-model Vision Transformer Based on Feature Fusion for Classification of DFUC2021 Challenge | Publicación