add boolTofloat func for cast operator

mindspore/lite/nnacl/fp32/cast.c

@@ -17,6 +17,12 @@
 #include "nnacl/fp32/cast.h"
 #include "nnacl/fp32/common_func.h"
 
+void BoolToFloat32(const bool *input, float *output, int number) {
+  for (int i = 0; i < number; ++i) {
+    output[i] = (float)input[i];
+  }
+}
+
 void Uint8ToFloat32(const uint8_t *input, float *output, int number) {
   for (int i = 0; i < number; ++i) {
     output[i] = (float)input[i];

mindspore/lite/nnacl/fp32/cast.h

@@ -31,6 +31,7 @@ typedef struct CastParameter {
 #ifdef __cplusplus
 extern "C" {
 #endif
+void BoolToFloat32(const bool *input, float *output, int number);
 void Uint8ToFloat32(const uint8_t *input, float *output, int number);
 void Uint8ToInt8(const uint8_t *input, int8_t *output, int number);
 void Int8ToUint8(const int8_t *input, uint8_t *output, int number);

mindspore/lite/src/runtime/kernel/arm/fp32/cast.cc

@@ -82,6 +82,9 @@ int CastCPUKernel::DoCast(int thread_id) {
     }
   } else {
     switch (input_data_type) {
+      case kNumberTypeBool:
+        BoolToFloat32(reinterpret_cast<bool *>(input->MutableData()) + offset,
+                      reinterpret_cast<float *>(output_data) + offset, data_num);
       case kNumberTypeUInt8:
         Uint8ToFloat32(reinterpret_cast<uint8_t *>(input->MutableData()) + offset,
                        reinterpret_cast<float *>(output_data) + offset, data_num);

mindspore/lite/tools/converter/parser/tflite/tflite_cast_parser.cc

@@ -46,9 +46,6 @@ STATUS TfliteCastParser::Parse(TfliteTensorsInfo *tensors_info, const std::uniqu
     return RET_NULL_PTR;
   }
   attr->srcT = GetTfliteDataType(in_tensor->type);
-  if (attr->srcT == TypeId::kNumberTypeBool) {
-    attr->srcT = TypeId::kNumberTypeUInt8;
-  }
   const auto &out_tensor = tflite_model->subgraphs[0]->tensors[tflite_op->outputs[0]];
   if (out_tensor == nullptr) {
     MS_LOG(ERROR) << "tensor is null";