Loading trained weights for batch normalization layer in Scale Layer

Hello,

I am trying to create Resnet50 model using Jetson c++ API. I have trained resnet50 on custom dataset using tensorflow.

STEP 1:
Model --> convert weights to .wts (x86 machine)
STEP 2:
loading .wts file --> cpp code handling stuff (Jetson x2)

// implementation way
Weights power{DataType::kFLOAT, nullptr, 0};
Weights shift{DataType::kFLOAT, &beta.values, beta.count};
Weights scale{DataType::kFLOAT, &gamma.values, gamma.count};
auto bn_layer1 = network->addScale(*conv1->getOutput(0), ScaleMode::kUNIFORM, scale, shift, power);
assert(bn_layer1 != nullptr);

I don’t know if this right way, can some one please the right way to pass values. for example, my gamma has shape (7, 7, 3, 64), so total count = 9048 but how do I pass value paramater.

Thanks