def get_data():
person_name = ["bear", "rabbit", "haha"]
dirname = "jsondata"
epochs = 50
offset = "none"
data = load(dirname + '/bear_data.npy')
labels = load(dirname + '/bear_labels.npy')
c = np.array(np.full(data.shape[0], 0))
labels = np.vstack((labels, c))
for i in range(2):
a = load(dirname + '/' + person_name[i + 1] + '_data.npy')
b = load(dirname + '/' + person_name[i + 1] + '_labels.npy')
c = np.array(np.full(a.shape[0], i + 1))
b = np.vstack((b, c))
data = np.vstack((data, a))
labels = np.hstack((labels, b))
a = data
print(labels)
# print(pre.make_velocity(data).shape)
# data = np.concatenate((data,pre.make_offset(a)),axis=2)
data = np.concatenate((data, pre.make_velocity(a)), axis=2)
# data = np.concatenate((data,pre.make_accel(a)),axis=2)
# print(data.shape)
data_test = data[np.where(labels[0, :] == 5), :, :, :]
data = data[np.where(labels[0, :] <= 2), :, :, :]
data_test = data_test.reshape(data_test.shape[1], data_test.shape[2],
data_test.shape[3] * data_test.shape[4])
data = data.reshape(data.shape[1], data.shape[2],
data.shape[3] * data.shape[4])
print(data.shape)
# labels = keras.utils.to_categorical(labels)
# print(labels)
labels_test = labels[:, np.where(labels[0, :] == 5)]
labels = labels[:, np.where(labels[0, :] <= 2)]
labels_test = labels_test.reshape(labels_test.shape[0],
labels_test.shape[2])
labels = labels.reshape(labels.shape[0], labels.shape[2])
index = np.arange(data.shape[0])
np.random.shuffle(index)
data = data[index]
labels[0] = labels[0][index]
labels[1] = labels[1][index]
# print(keras.utils.to_categorical(labels[0]))
# print(keras.utils.to_categorical(labels[1]))
labels_v = keras.utils.to_categorical(labels[0], num_classes=6)
labels = keras.utils.to_categorical(labels[1], num_classes=3)
print(labels_v)
print(labels)
labels_test = keras.utils.to_categorical(labels_test[1], num_classes=3)
return data, data_test, labels, labels_test, labels_v
plt.scatter(X_norm[i, 0], X_norm[i, 1], c=cValue[y[i]], alpha=0.7)
plt.show()
if __name__ == '__main__':
filename = 'gym_6kind_012_5_'
action_start = 0
X_train = load(filename + "data.npy")
Y_train = load(filename + "labels.npy")
Y_p_train = load(filename + "labels_p.npy")
X_test = load(filename + "data_test.npy")
Y_test = load(filename + "labels_test.npy")
Y_p_test = load(filename + "labels_p_test.npy")
tmp = pre.make_velocity(X_train)
X_train = np.concatenate((X_train, tmp), axis=-1)
tmp = pre.make_velocity(X_test)
X_test = np.concatenate((X_test, tmp), axis=-1)
#X_train = pre.make_velocity(X_train)
#X_test = pre.make_velocity(X_test)
GAN0 = distanglingGAN()
GAN0.train(X_train,
Y_train,
Y_p_train,
X_test,
Y_test,
Y_p_test,
action_start,
n_batch=3000,
'./data/json_data/action/json_data_seperate/json_data_seperate_senior/data.npy'
)
#angle = pre.make_angle(data)
data_test = load(
'./data/json_data/action/json_data_seperate/json_data_seperate_senior/data_test.npy'
)
#angle_test = pre.make_angle(data_test)
labels = load(
'./data/json_data/action/json_data_seperate/json_data_seperate_senior/labels.npy'
)
labels_test = load(
'./data/json_data/action/json_data_seperate/json_data_seperate_senior/labels_test.npy'
)
#data = pre.make_offset(data)
#data_test = pre.make_offset(data_test)
data = pre.make_velocity(data)
data_test = pre.make_velocity(data_test)
index = np.arange(data.shape[0])
np.random.shuffle(index)
data = data[index]
labels = labels[index]
data = data.reshape(data.shape[0], data.shape[1],
data.shape[3] * data.shape[2])
data_test = data_test.reshape(data_test.shape[0], data_test.shape[1],
data_test.shape[3] * data_test.shape[2])
#mix_data = mix_data.reshape(mix_data.shape[0],mix_data.shape[1],mix_data.shape[2]*mix_data.shape[3])
# data /= 3000
# data_test /= 3000
# data,data_test,labels,labels_test = train_test_split(data,labels,test_size = 0.2,random_state = 10)