在 tflite model maker 上使用 mnist 数据集

Question

I would like to try tflite model maker which provides pretrained machine learning models for softwares. I'm new at it and i'd like to set a tflite model with "mnist" dataset for image classification api. I took my data from keras.datasets but i can't import it with "DataLoader" which is in tflite.

My question: Is there any way to use array data on image_classifier?

my codes:

from tflite_model_maker import image_classifier
from tflite_model_maker.image_classifier import DataLoader
from sklearn.datasets import load_digits

data = load_digits()

# Load input data specific to an on-device ML app.
train_data, test_data = data.split(0.9)

# Customize the TensorFlow model.
model = image_classifier.create(train_data)

# Evaluate the model.
loss, accuracy = model.evaluate(test_data)

Error i encountered:

KeyError: 'split'

During handling of the above exception, another exception occurred:

AttributeError                            Traceback (most recent call last)
1 frames
/usr/local/lib/python3.7/dist-packages/sklearn/utils/__init__.py in __getattr__(self, key)
    117             return self[key]
    118         except KeyError:
--> 119             raise AttributeError(key)
    120 
    121     def __setstate__(self, state):

AttributeError: split

Answer 1

load_digits() object has no attribute split, use train_test_split from sklearn. For example:

train_data, test_data = train_test_split(data, target, test_size=0.01)