Evaluate_Model

src.Evaluate_Model.evaluate_tflite_model(tflite_model, tfl_int8=True)

Evaluate the performance of a TFLite model.

Parameters

tflite_modelstr

Path to the TFLite model.

tfl_int8bool, optional

If True, use int8 quantized TFLite model. If False, use float32 TFLite model.

Returns

tflite_accuracyfloat

Accuracy of the TFLite model on the test dataset.

inference_speedfloat

Average inference speed.

src.Evaluate_Model.model_evaluation(trained_model, test_ds)

Evaluate the model performance.

Parameters

trained_modelkeras.Model

The trained model to be evaluated.

test_dstensorflow.data.Dataset

Test dataset.

Returns

raw_model_accuracyfloat

Model accuracy on the test dataset.