TfLite Android: Garbage values when running inference for multiple output model

Question

I have a model that predicts the age and gender of the input image of size 160X160. I am creating a byte buffer to input the image to the model and everything works just fine when using a model with only one output.

But when I am using the tflite.runForMultipleInputsOutputs(), I am getting garbage values which are of the form -> [[F@e233 etc.

I have followed the documentation and the sample apps to the detail and have been stuck at this for almost 2 days. Please help.

I am posting my code below for reference.

The model has 2 outputs:

Edit:

age -> float32 [1, 101]

gender -> float32 [1,2]

P.S - I am not doing anything with the output as of now. I just want to see the result of the model.

String classifyImage(Bitmap bitmap){
        try{
            ByteBuffer byteBuffer = convertBitmaptoByteBuffer(bitmap);

            float[][] out_gender = new float[1][2];
            float[][] out_age = new float[1][101];
            Object[] input = {byteBuffer};

            Map<Integer, Object> outputs = new HashMap();
            outputs.put(0, out_age);
            outputs.put(1, out_gender);

            interpreter.runForMultipleInputsOutputs(input, outputs);
            
        }catch (Exception e){
            e.printStackTrace();
        }
        return "";
    }

Joel · Accepted Answer · 2020-09-01 08:09:12Z

2

First, I would suggest that you double-check that the outputs from your model match your output map. It seems strange to me that the gender would be a 101-dimensional array and the age a 2-dimensional one. Have you by any chance mixed those up?

Secondly, I think you are calling toString() on the float arrays. Consider using e.g. System.out.println(Arrays.deepToString(out_age)); to present the result.

answered Sep 1, 2020 at 8:09

Joel

481 silver badge10 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

abhishekti7 Over a year ago

Thank you for the suggestion @Joel. the age and gender shape was my mistake in question. I have corrected it. I will try the string suggestion and let you know

Collectives™ on Stack Overflow

TfLite Android: Garbage values when running inference for multiple output model

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related