ValueError: Error when checking input: expected dense_1_input to have shape (9,) but got array with shape (1,)

Question

hi so I build a DNN network to classify some objects in an image using the features of the object, like bellow :

contours, _ = cv2.findContours(imgthresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE)

for contour in contours:
    features = np.array([])
    (x_start, y_start, character_width, character_height) = cv2.boundingRect(contour)
    x_end = x_start + character_width
    y_end = y_start + character_height
    character_area = character_width * character_height
    features = np.append(features, [character_width, character_height, character_area, x_start,
                                    y_start, x_end, y_end, image_width, image_height])

    print(features)
    print(features.shape)
    cv2.rectangle(image, (x_start, y_start), (x_end, y_end), (0, 255, 0), thickness=1)

print(features) output is:

[  5.   1.   5. 105.  99. 110. 100. 100. 117.]

and print(features.shape) is:

(9,)

I build and trained a DNN using the following code :

model = Sequential()

model.add(Dense(50, input_dim=9, activation='relu'))

model.add(Dropout(0.5))

model.add(Dense(40, activation='relu'))

model.add(Dropout(0.5))

model.add(Dense(30,activation='relu'))

model.add(Dense(2, activation='softmax'))

The input layer has 9 input features. So I tried to get the prediction of the model using:

model.predict_classes(features)

The training data, a CSV file, contains 10 columns (9 features and 1 for the output)

I got the following error :

ValueError: Error when checking input: expected dense_1_input to have shape (9,) but got array with shape (1,)

I tried to reshape the features array by using :

np.reshape(features,(1,9)

but that didn't work either. I am still new at this field

Answer 1

Here is a minimal working example.

import numpy as np
import tensorflow as tf

def main():
    features = np.array([5, 1, 5, 105, 99, 110, 100, 100, 117])
    model = tf.keras.Sequential()
    model.add(tf.keras.layers.Dense(50, input_dim=9, activation="relu"))

    print(tf.expand_dims(features, 0))
    print(np.reshape(features, (1, 9)))

    print(model.predict_classes(np.reshape(features, (1, 9))))


if __name__ == '__main__':
    main()

As you can see, the np.reshape call make it works. It is roughly equivalent to the tf.expand_dims.

Your current error comes from the fact that your model expect a batch dimension. So, if you pass it an array of shape (9,) it infers that it's a batch of scalars, and not a single array of size 9.