I'm having trouble deploying the best_float16.tflite model to Kotlin because it has a different data type compared to the code I'm following. Here are the differences I have identified...
As you can see, there are differences in the output between my model and the other model. (pict 1 : my model : best_float16.tflite, pict 2: another model) I'm confused about how to call it like the code below:
val outputs = model.process(image)
val locations = outputs.locationsAsTensorBuffer.floatArray
val classes = outputs.classesAsTensorBuffer.floatArray
val scores = outputs.scoresAsTensorBuffer.floatArray
val numberOfDetections = outputs.numberOfDetectionsAsTensorBuffer.floatArray
var mutable = bitmap.copy(Bitmap.Config.ARGB_8888, true)
val canvas = Canvas(mutable)
val h = mutable.height
val w = mutable.width
paint.textSize = h/15f
paint.strokeWidth = h/85f
var x = 0
scores.forEachIndexed { index, fl ->
x = index
x *= 4
if(fl > 0.5){
paint.setColor(colors.get(index))
paint.style = Paint.Style.STROKE
canvas.drawRect(RectF(locations.get(x+1)*w, locations.get(x)*h, locations.get(x+3)*w, locations.get(x+2)*h), paint)
paint.style = Paint.Style.FILL
canvas.drawText(labels.get(classes.get(index).toInt())+" "+fl.toString(), locations.get(x+1)*w, locations.get(x)*h, paint)
}
"I tried to retrieve classes, locations, and scores, but they are not available, and using TensorBuffer also doesn't work. I can only use outputAsCategoryList like below:
I need advice from those of you who know how to implement the best_float16.tflite model and suggestions on which part of my code is incorrect and needs to be fixed.
