TensorFlow in iOS Swift:Binary classification problem

TensorFlow General Discussion
, ,
1

I am having some issues while getting output in TensorFlow Model(tflite).

This is my input:

let inputInfo:[Float32]  : [-1.0291401 ,  1.6121695 ,  0.58366895, -0.25974554,  2.6633718 ,
                                  0.39398468,  1.2648116 , -1.0617405 ,  1.0997621 , -0.01813432,
                                 -0.02543107,  1.9113901 ,  0.30188444,  0.3199759 ,  0.07759953,
                                  0.23082322,  2.0959156 , -0.42658705,  0.08775132,  3.4258583 ,
                                 -1.0573974 ,  0.7249298 , -1.1119401 , -0.72663903, -0.74873704,
                                 -0.387724  , -0.14288527, -0.39554232, -0.10774904, -0.0911286 ,
                                  0.40389383, -0.169619  , -1.1736624 ]

 let inputData = Data(bytes: &inputInfo, count: inputInfo.count * MemoryLayout<Float32>.stride)
 print(inputData)
 try interpreter?.copy(inputData, toInputAt: 0)

After passing this array of Float32 type as input I am getting below result in output:

Output:

TensorFlowLite.Tensor(name: “Identity”, dataType: TensorFlowLite.Tensor.DataType.float32, shape: TensorFlowLite.Tensor.Shape(rank: 2, dimensions: [1, 1]), data: 4 bytes, quantizationParameters: nil)

For getting the expexcted output result I am using this code:

let outputTensor = try self.interpreter?.output(at: 0)
let finalOutput = [Float32](unsafeData: outputTensor!.data)
 print(finalOutput)
 4.6533377e+33

Here expected output should be some numbers between 0 & 1(like 0.8,0.9) but my final output is beyond that expected output. I am stuck here,please help.

3

Hi, @Priyanshu_Srivastava

I apologize for the delayed response, I’m not sure whether you’re doing image classification or text classification so could you please refer these TensorFlow Lite text classification iOS example and image classification iOS example which may help you to solve your issue ?

If issue still persists please help us with your Github repo (If possible).

Thank you for your cooperation and patience.