Fix validation error in graph_ssd_mobilenet
The graph example has fixed quantization information given for certain layers, and some of the offsets exceed the 8-bit range for Int8 data type.
This shouldn't have been the case and the offsets should respect the 8-bit quantization specification laid out here: https://www.tensorflow.org/lite/performance/quantization_spec
However, the mechanism added in the helper function introduces robustness in case of such irregularities with little/no cost; and therefore added as a fix.
Resolves: COMPMID-6748
Change-Id: If39bf323382f109fa100ee2b87ce63cc7bc89759
Signed-off-by: Gunes Bayir <gunes.bayir@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/10858
Reviewed-by: SiCong Li <sicong.li@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
diff --git a/src/core/Utils.cpp b/src/core/Utils.cpp
index 90a7ac3..532d08d 100644
--- a/src/core/Utils.cpp
+++ b/src/core/Utils.cpp
@@ -450,8 +450,9 @@
const int b_int = is_qasymm8_signed ? quantize_qasymm8_signed(b, oq_info) : quantize_qasymm8(b, oq_info);
const auto type_max_value = std::get<1>(get_min_max(data_type)).get<int32_t>();
- const int32_t min_activation =
- act_info.activation() != ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU ? oq_info.offset : b_int;
+ const int32_t min_activation = act_info.activation() != ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU
+ ? std::min(oq_info.offset, type_max_value)
+ : b_int;
const int32_t max_activation =
act_info.activation() == ActivationLayerInfo::ActivationFunction::RELU ? type_max_value : a_int;