0

Is it possible to use the ML kit custom model hosting to do object detection not just object classification. I am familiar with doing this in the tensorflow lite sample android application, yet I would like to use Firebase as it has capability of updating model and iOS/android support.

Basically, I am not sure how to this beginning with the line:

int[] outputDims = {DIM_BATCH_SIZE,Not sure what other values since I am not doing just classification };

Does the firebase model interpreter support this?

radish25
  • 75
  • 7

1 Answers1

0

Yes, the ML Kit custom model interpreter should support your object detection TFLite models. The input/output dimensions in the code should match your TFLite model input/output.

Since you are familiar with TFLite Android sample app. Let's take TFLite Android sample app and ML Kit sample app as an example here.

In TFLite Android sample app, we define the output dimensions as two-dim array with float type: https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/java/demo/app/src/main/java/com/example/android/tflitecamerademo/ImageClassifierFloatInception.java#L38

In ML Kit sample app, we also define it as two dimension array with float type: https://github.com/firebase/quickstart-android/blob/master/mlkit/app/src/main/java/com/google/firebase/samples/apps/mlkit/java/custommodel/CustomImageClassifier.java#L191

The dimensions are expressed as a list, and the data type is a parameter in ML Kit.

Shiyu
  • 875
  • 4
  • 5