EFormer: Enhanced Transformer towards Semantic-Contour Features of Foreground for Portraits Matting | Read Paper on Bytez