Incorrect predictions (tensorflow model converted to core ml)

I converted my model with coremltools:

model = tf.keras.models.load_model('mymodel')

mlmodel = ct.convert(model, convert_to="mlprogram",inputs=[ct.ImageType( name="sequential_1_input", shape=(1, 192, 192, 3),

scale=1.0 / 255.0, bias=[-10,-10,-10], color_layout="RGB"], source="tensorflow", classifier_config=ct.ClassifierConfig(class_labels) )

Set model author name

mlmodel.author = 'name'

Set the license of the model

mlmodel.license ="license"

Set a short description for the Xcode UI

mlmodel.short_description ="desc "

Set a version for the model

mlmodel.version = "1.0"

mlmodel.save("my model.mlpackage") `

Then use this on my iOS app

      let buffer = unwrapImage.buffer(with: CGSize(width:192, height:192)) # making sure the image is 192x192


      let config  = MLModelConfiguration()
      let imageClassifier = try MyModel(configuration: config)
      let result = try imageClassifier.prediction(sequential_1_input: buffer!)



      let className: String = result.classLabel
      let confidence: Double = result.classLabel_probs[result.classLabel] ?? 1.0
      self.the_text = ("\(className)\nWith Confidence:\n\(confidence)")

For testing prediction on Core ML I'm using exactly the same image I'm using in python, but the predictions are all the time wrong, not even close!

So I'm guessing I'm doing something wrong, either when I do the conversion or after, any ideas?