Tensorflow Convolucional Neural Networks

Question

Hey, you guys, hey! I have a problem, I believe, in my training set. I have a numpy array with 649 images for training my network however when starting the training(model.fit) I see that only 21 images are "coming" for this training. Someone would know to tell me what might be going on?

model = tf.keras.models.Sequential()
model.add(tf.keras.layers.Conv2D(filters=32, kernel_size=3, padding="same", activation="relu", input_shape=[100, 100, 3]))
model.add(tf.keras.layers.Conv2D(filters=32, kernel_size=3, padding="same", activation="relu"))
model.add(tf.keras.layers.MaxPool2D(pool_size=2, strides=2, padding='valid'))
model.add(tf.keras.layers.Conv2D(filters=64, kernel_size=3, padding="same", activation="relu"))
model.add(tf.keras.layers.Conv2D(filters=64, kernel_size=3, padding="same", activation="relu"))
model.add(tf.keras.layers.MaxPool2D(pool_size=2, strides=2, padding='valid'))
model.add(tf.keras.layers.Flatten())
model.add(tf.keras.layers.Dense(units=128, activation='relu'))
model.add(tf.keras.layers.Dropout(0.2))
model.add(tf.keras.layers.Dense(units=128, activation='relu'))
model.add(tf.keras.layers.Dropout(0.2))
model.summary()
model.compile(loss="sparse_categorical_crossentropy",
              optimizer="Adam", metrics=["sparse_categorical_accuracy"])
model.fit(X_train, y_train, epochs=25)

Answer 1

The function .fit() has as a standard a batch_size=32. This means that for a weight upgrade (backpropagation), 32 images will be used at a time. As your dataset has 649 images, will be realziados 21 backpropagations per time. (649/32 = 20.3)

If you want 649 backpropagations per season, use batch_size=1.