How to run trained mxnet model in Java

[anjishnu]

I want to develop and experiment with models in Python but execute them in Java (e.g. in an Android App).

How do I accomplish something analogous to this -
https://medium.com/google-cloud/how-to-invoke-a-trained-tensorflow-model-from-java-programs-27ed5f4f502d

[thirdwing]

Can you try this https://github.com/dmlc/mxnet/tree/master/amalgamation#android?

[zihaolucky]

If not in Android, you can build/make a MXNet-Scala package and put the jar into your project, then write a predictor.scala to accept java array input and serve.

class MXNetPredictor (prefix: String, epoch: Int, batchSize: Int) {

  val model = FeedForward.load(prefix, epoch)

  /**
    * predict
    *
    * @param flat A flat feature input vector.
    * @param shape Shape of input data.
    * @return
    */
  def predict(flat: Array[Float], shape: Array[Int]): Array[NDArray] = {
    val ndArray = NDArray.array(flat, Shape(shape))

    val data: IndexedSeq[NDArray] = IndexedSeq(ndArray)
    val label: IndexedSeq[NDArray] = IndexedSeq()

    val valData: NDArrayIter = new NDArrayIter(data, label, batchSize)
    val prediction = model.predict(valData)
    return prediction

  }

  /**
    * Top-1 prediction results for a batch data.
    *
    * @param flat A flat feature input vector.
    * @param shape Shape of input data.
    * @return Return top-1 prediction results for a batch data.
    */
  def predictTop1(flat: Array[Float], shape: Array[Int]): NDArray = {
    val prediction = predict(flat, shape)(0)
    NDArray.argmaxChannel(prediction)
  }

}

In Java side, it's simple

private MXNetPredictor          predictor;

[yajiedesign]

This issue is closed due to lack of activity in the last 90 days. Feel free to reopen if this is still an active issue. Thanks!

[sampathchanda]

@zihaolucky Do you have a git repo or some working example of the same? I am new to both java and scala but trying to deploy an MXNet model using Java. It would be great if you can point me to a working example of the same.