Feeding an image into an already trained TensorFlow CNN

Question

I am somewhat new to Python and very new to TensorFlow. I have followed a couple tutorials and I am stuck on this colab with this video . Training ran perfectly and I have a saved model. Now I want to load that model and feed one of my own images into it. Here is my attempt:

import tensorflow as tf
import cv2
from tensorflow import keras
model = tf.keras.models.load_model('rps.h5')
model.summary()

img = cv2.imread('my_hand_paper.png')
print(model.predict_classes(img))

but I get the following error:

ValueError: Input 0 of layer sequential is incompatible with the layer: expected ndim=4, found ndim=3. Full shape received: [None, 300, 3]

My image is 300x300 just like the training images. I imagine the problem is that I have to prepare the image in a similar way to the training data, but I am unsure how. This is how the training data was prepared:

training_datagen = ImageDataGenerator(
  rescale = 1./255,
  rotation_range=40,
  width_shift_range=0.2,
  height_shift_range=0.2,
  shear_range=0.2,
  zoom_range=0.2,
  horizontal_flip=True,
  fill_mode='nearest')

train_generator = training_datagen.flow_from_directory(
  TRAINING_DIR,
  target_size=(150,150),
  class_mode='categorical',
  batch_size=126)

Output from summary():

 Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d (Conv2D)              (None, 148, 148, 64)      1792      
_________________________________________________________________
max_pooling2d (MaxPooling2D) (None, 74, 74, 64)        0         
_________________________________________________________________
conv2d_1 (Conv2D)            (None, 72, 72, 64)        36928     
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 36, 36, 64)        0         
_________________________________________________________________
conv2d_2 (Conv2D)            (None, 34, 34, 128)       73856     
_________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 17, 17, 128)       0         
_________________________________________________________________
conv2d_3 (Conv2D)            (None, 15, 15, 128)       147584    
_________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 7, 7, 128)         0         
_________________________________________________________________
flatten (Flatten)            (None, 6272)              0         
_________________________________________________________________
dropout (Dropout)            (None, 6272)              0         
_________________________________________________________________
dense (Dense)                (None, 512)               3211776   
_________________________________________________________________
dense_1 (Dense)              (None, 3)                 1539      
=================================================================
Total params: 3,473,475
Trainable params: 3,473,475
Non-trainable params: 0
_________________________________________________________________

Answer 1

When you want to use your network in inference, you still have to use a batch size. In your case, the batch size is 1.

You can add the batch with the following code:

img = cv2.resize(img, (150,150))    
img = tf.expand_dims(img , 0)