已合并
fix: 从 ops-cv 迁出 NormalizeBBox 原型 #1456
fix: 从 ops-cv 迁出 NormalizeBBox 原型 #1456
已合并
raoliang_sac创建于 9月2日
共 1 个文件变更+0-21
@@ -985,27 +985,6 @@ REG_OP(NonMaxSuppressionV7)
985 .ATTR(max_boxes_size, Int, 0)985 .ATTR(max_boxes_size, Int, 0)
986 .OP_END_FACTORY_REG(NonMaxSuppressionV7)986 .OP_END_FACTORY_REG(NonMaxSuppressionV7)
987 987 
988-/**
989- *@brief Computes Normalize bbox function.
990- *
991- *@par Inputs:
992- *Inputs include:
993- * @li boxes: A Tensor. Must be float16 or float32.
994- * @li shape_hw: A Tensor. Must be int32.
995- *
996- *@par Attributes:
997- * reversed_box: optional, bool. Defaults to "False"
998- *
999- *@par Outputs:
1000- * y: A Tensor. Must have the same type and shape as boxes.
1001- */
1002-REG_OP(NormalizeBBox)
1003- .INPUT(boxes, TensorType({DT_FLOAT16, DT_FLOAT}))
1004- .INPUT(shape_hw, TensorType({DT_INT32}))
1005- .OUTPUT(y, TensorType({DT_FLOAT16, DT_FLOAT}))
1006- .ATTR(reversed_box, Bool, false)
1007- .OP_END_FACTORY_REG(NormalizeBBox)
1008- 
1009/**988/**
1010* @brief Performs the backpropagation of ROIAlign for training scenarios . \n989* @brief Performs the backpropagation of ROIAlign for training scenarios . \n
1011 990