return K.mean(K.square((difference * weights)), axis=-1)
inputLayer = Input(shape=(21, ))
hiddenLayer1 = Dense(11)(inputLayer)
outputLayer = Dense(7)(hiddenLayer1)
model = Model(inputs=inputLayer, outputs=outputLayer)
model.compile(optimizer='adam',
loss='mean_squared_error',
metrics=['accuracy'])
p = Parser()
dataFileTrain = sys.argv[1]
dataFileTest = sys.argv[2]
inputDataTrain = array(p.Parse(dataFileTrain))
print(inputDataTrain.shape)
outputDataTrain = array(p.ParseSpine(dataFileTrain))
print(outputDataTrain.shape)
history = model.fit(inputDataTrain, outputDataTrain, 32, 2000)
print(model.outputs)
[print(n.name) for n in K.get_session().graph.as_graph_def().node]
# summarize history for loss
plt.plot(history.history['loss'])
plt.title('model loss')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train', 'test'], loc='upper left')
plt.show()
# tf.train.Saver().save(K.get_session(), export_path + '/checkpoint.ckpt')
# tf.train.write_graph(K.get_session().graph.as_graph_def(),
# export_path, 'graph.pbtxt', as_text=True)
# tf.train.write_graph(K.get_session().graph.as_graph_def(),
# export_path, 'graph.pb', as_text=False)
# freeze_graph.freeze_graph(input_graph = export_path +'/graph.pbtxt',
# input_binary = False,
# input_checkpoint = export_path + '/checkpoint.ckpt',
# output_node_names = "dense_2/BiasAdd",
# output_graph = export_path +'/model.bytes' ,
# clear_devices = True, initializer_nodes = "",input_saver = "",
# restore_op_name = "save/restore_all", filename_tensor_name = "save/Const:0")
inputDataTest = array(p.Parse(dataFileTest))
outputDataTest = array(p.ParseSpineRotation(dataFileTest))
#loss_and_metrics = model.evaluate(inputDataTest, outputDataTest)
#print(loss_and_metrics)
#test the model
test = []
tolist = list(inputDataTest[0])
e = Estimator()
estimate = e.Estimate([inputDataTest[0]])
print(estimate[0][0])
print(estimate[0][1])
print(estimate[0][2])
tolist.append(
estimate[0]
[0]) ###########################use estimator to get first frame prev.