使用 Keras 和 Tensorflow 的 model.predict 问题

Question

Here i give a test_input of 28 cols and +1000 row the trained model should accept it but i'm getting an error of shape compatibility

here is my code:

my_data = genfromtxt('test_rgb.csv', delimiter=',',skip_header=1)
test_data=my_data[0:,1:-1]

for test_row in test_data:
    predictions = model.predict(test_row)
    print(predictions)

Answer 1

When you use predict you must supply your model with tensor (list of vectors) which means that if you want to make a prediction from only one piece of data you must reshepe it like (1,(INPUT_SHAPE))

in your case try to use if

model.predict(np.array(test_row).reshape((1,a,b,c)))

you specified input_shape=(a,b,c,)

Answer 2

It says right in the error, the input shape your model needs (28,), the data you are providing has a shape of (1,)

You can try printing the shape of your data prior to your predict method to verify. You probably just need to reshape your data.

Answer 3

Thanks all, the solution is to reshape the row_test

test_row=test_row.reshape(1,28)