training a convnet to predict various scenes from the Kaggel Intel Image dataset
Training on gcloud archtitecture Current accuracy = 53.4% (in progress)
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
conv2d_1 (Conv2D) (None, 150, 150, 72) 2016
_________________________________________________________________
conv2d_2 (Conv2D) (None, 148, 148, 64) 41536
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 29, 29, 64) 0
_________________________________________________________________
conv2d_3 (Conv2D) (None, 29, 29, 64) 36928
_________________________________________________________________
conv2d_4 (Conv2D) (None, 27, 27, 48) 27696
_________________________________________________________________
conv2d_5 (Conv2D) (None, 25, 25, 32) 13856
_________________________________________________________________
conv2d_6 (Conv2D) (None, 23, 23, 24) 6936
_________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 4, 4, 24) 0
_________________________________________________________________
flatten_1 (Flatten) (None, 384) 0
_________________________________________________________________
dense_1 (Dense) (None, 128) 49280
_________________________________________________________________
dense_2 (Dense) (None, 96) 12384
_________________________________________________________________
dense_3 (Dense) (None, 64) 6208
_________________________________________________________________
dropout_1 (Dropout) (None, 64) 0
_________________________________________________________________
dense_4 (Dense) (None, 6) 390
=================================================================
Total params: 197,230
Trainable params: 197,230
Non-trainable params: 0