diff options
author | FinnWilliamsArm <Finn.Williams@arm.com> | 2019-05-22 14:50:55 +0100 |
---|---|---|
committer | Narumol Prangnawarat <narumol.prangnawarat@arm.com> | 2019-05-27 16:27:36 +0000 |
commit | af8b72d3a97baa0c86583b03321e8a8d3f4bc2aa (patch) | |
tree | 3b4d86842399842b852e593c3bcf131d86d3ae20 /tests/InferenceTestImage.hpp | |
parent | 43aec5886449c1b024b740fd6f4500e827bde221 (diff) | |
download | armnn-af8b72d3a97baa0c86583b03321e8a8d3f4bc2aa.tar.gz |
IVGCVSW-3129 Image pre-processing fix for TFLite
* Resized images for quantized models are now statically cast to uint8
instead of quantized
* Removed optional quantization parameters from ImagePreprocessor
constructor
* Changed mean and scale for TFLite models
Signed-off-by: FinnWilliamsArm <Finn.Williams@arm.com>
Change-Id: Id5ffdf77f3614d10c417e769bd8ffc4a4c07308b
Diffstat (limited to 'tests/InferenceTestImage.hpp')
-rw-r--r-- | tests/InferenceTestImage.hpp | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/tests/InferenceTestImage.hpp b/tests/InferenceTestImage.hpp index 59f4cc7073..643d060919 100644 --- a/tests/InferenceTestImage.hpp +++ b/tests/InferenceTestImage.hpp @@ -92,7 +92,8 @@ public: const armnn::CheckLocation& location, const ResizingMethods meth = ResizingMethods::STB, const std::array<float, 3>& mean = {{0.0, 0.0, 0.0}}, - const std::array<float, 3>& stddev = {{1.0, 1.0, 1.0}}); + const std::array<float, 3>& stddev = {{1.0, 1.0, 1.0}}, + const float scale = 255.0f); void Write(WriteFormat format, const char* filePath) const; |