Then, an attention module recurrently generates a vertical weighted mask enabling to focus on the current text line features.
確定! 回上一頁