According to the characteristics of face mask data set, by defining a new distance formula and using k-means algorithm to re cluster anchors, a more suitable prior frame size is generated and assigned to the corresponding prediction layer to solve the imprecise problem of frame regression.<br>
正在翻译中..