from datahelper import DataHelper
x = SimpleNet.x
cnn = SimpleNet.build_cnn(x, reuse=None, raw_out=True)
rnn = SimpleRNN.build_graph(cnn)
infer = tf.nn.softmax(rnn)
y = SimpleRNN.y
dh = DataHelper()
saver = tf.train.Saver()
#print('Recovering')
#saver.recover_last_checkpoints('../models')
with tf.Session() as sess:
print('Restoring')
saver.restore(sess, '../models/cnn-rnn-100.ckpt')
while True:
img, lbl = dh.get_next_example(split_channels=True)
_y = sess.run(infer, feed_dict={x:img})
print(f'Label:{lbl}\nGuess:{_y}')
for i in range(4):
plt.figure()
plt.imshow(img[i,:,:,0], cmap='gray')
plt.show()
