MLECO-2599: Replace DSCNN with MicroNet for KWS
Added SoftMax function to Mathutils to allow MicroNet
to output probability as it does not nativelu have this layer.
Minor refactoring to accommodate Softmax Calculations
Extensive renaming and updating of documentation and resource download script.
Added SoftMax function to Mathutils to allow MicroNet
to output probability.
Change-Id: I7cbbda1024d14b85c9ac1beea7ca8fbffd0b6eb5
Signed-off-by: Liam Barry <liam.barry@arm.com>
diff --git a/source/use_case/asr/include/AsrClassifier.hpp b/source/use_case/asr/include/AsrClassifier.hpp
index 2c97a39..67a200e 100644
--- a/source/use_case/asr/include/AsrClassifier.hpp
+++ b/source/use_case/asr/include/AsrClassifier.hpp
@@ -32,12 +32,13 @@
* populated by this function.
* @param[in] labels Labels vector to match classified classes
* @param[in] topNCount Number of top classifications to pick.
+ * @param[in] use_softmax Whether softmax scaling should be applied to model output.
* @return true if successful, false otherwise.
**/
bool GetClassificationResults(
TfLiteTensor* outputTensor,
std::vector<ClassificationResult>& vecResults,
- const std::vector <std::string>& labels, uint32_t topNCount) override;
+ const std::vector <std::string>& labels, uint32_t topNCount, bool use_softmax = false) override;
private:
/**