def __init__(self):
self.Kp = 6.0 #7.0
self.Ki = 0.5 #2.0
self.Kd = 3.0 #3.0
self.mode = 0
self.errorL_1 = 0.0
self.UiL_1 = 0.0
self.errorR_1 = 0.0
self.UiR_1 = 0.0
self.ref_WL = 0.0
self.ref_WR = 0.0
self.wr = 0.0
self.wl = 0.
self.il = 0.0
self.ir = 0.0
self.encl = 0.0
self.encr = 0.0
self.a = np.arry([0, 0, 0])
self.av = np.arry([0, 0, 0])
topics_list = [
'/UL', '/UR', '/WL', '/WL_ref', '/WR', '/WR_ref', '/enc_L',
'/enc_R', '/iL', '/iR', '/voltage', '/um7', '/imu/data'
]
self.topic2save = ' '.join(topics_list)
(self.options, self.args) = parser.parse_args()
rospy.init_node("Ident_model")
def train(self,inputs_list,target_list):
#将输入列表转化为二阶矩阵
inputs = numpy.array(input_list,ndmin = 2).T
targets = numpy.arry(target_list,ndmin = 2).T
#如何进行隐藏层计算，同query函数中的一样
hidden_inputs = numpy.dot(self,wih,inputs)
hidden_inputs = self.activation_function(hidden_inputs)
#如何计算最终层，同上
final_inputs = numpy.dot(self.who,hidden_inputs)
final_outputs = self.activation_function(final_inputs)
#计算误差，误差 = 目标值target - 最终值final_outputs
output_errors = targets - final_outputs
#将输出时得到的误差根据权重反向分配到隐藏层，将误差进行重组
hidden_error = numpy.dot(self.who.T , output_errors)
#更新隐藏层和输出层之间的分配权重
self.who+= self.lr*numpy.dot((output_errors * final_outputs * (1.0 - final_outputs)),numpy.tranpose(hidden_outputs
))
#更新输入层和隐藏层之间的分配权重，方法同上
self.wih+= self.lr*numpy.dot((hidden_errors * hidden_outputs * (1.0 - hidden_outputs)),numpy.tranpose(inputs))
pass
def get_mean_std_dataset(root_dir):
"""
Function to compute mean and std of image dataset.
Move batch_size param according to memory resources.
retrieved from: https://forums.fast.ai/t/image-normalization-in-pytorch/7534/7
"""
# data_domain = "amazon"
# path_dataset = "datasets/office/%s/images" % data_domain
transform = transforms.Compose([
transforms.Resize((224, 224)), # original image size 300x300 pixels
transforms.ToTensor()
])
dataset = datasets.ImageFolder(root=root_dir, transform=transform)
# set large batch size to get good approximate of mean, std of full dataset
# batch_size: 4096, 2048
data_loader = DataLoader(dataset,
batch_size=2048,
shuffle=False,
num_workers=0)
mean = []
std = []
for i, data in enumerate(data_loader, 0):
# shape is (batch_size, channels, height, width)
npy_image = data[0].numpy()
# compute mean, std per batch shape (3,) three channels
batch_mean = np.mean(npy_image, axis=(0, 2, 3))
batch_std = np.std(npy_image, axis=(0, 2, 3))
mean.append(batch_mean)
std.append(batch_std)
# shape (num_iterations, 3) -> (mean across 0th axis) -> shape (3,)
mean = np.array(mean).mean(axis=0) # average over batch averages
std = np.arry(std).mean(axis=0) # average over batch stds
values = {"mean": mean, "std": std}
return values
def load_dataset(path, file_ext=".jpg"):
'''
resize the image
assuming row*col = 720*1080
'''
images = []
print ('------------------------------------------')
print ('loading data_set')
print ('path :' , path)
print ('file_ext : ', file_ext)
print ('------------------------------------------')
# images_list -> stores a list of names
# of images in the form of a strig
images_list = glob.glob(path + "*" + file_ext)
for i in images_list:
img = cv2.imread(i)
img = cv2.resize(img, (storage_class.resize_col, storage_class.resize_row), interpolation=cv2.INTER_LINEAR)
img = img.astype(np.float32)
images.append(img)
images = np.arry(images)
return images
def plot_return_risk():
ret, vol = return_risk(stocks)
color = np.arry([0.18, 0.96, 0.75, 0.3, 0.9, 0.5])
plt.scatter(ret,
vol,
marker='o',
c=color,
s=500,
camp=plt.get_cmap('Spectral'))
plt.xlable("日收益率均值%")
plt.ylable("标准差%")
for lable, x, y in zip(stocks.keys(), ret, vol):
plt.annotate(lable,
xy=(x, y),
xytext=(20, 20),
textcoords="offset points",
ha="right",
va="bottom",
bbox=dict(boxstyle='round,pad=0.5',
fc='yellow',
alpha=0.5),
arrowprops=dict(arrowstyle="->",
connetionstyle="arc3,rad=0"))
NameError: name 'x' is not defined
>>> X.shape
(26048, 72)
>>> X.describe
Traceback (most recent call last):
File "<pyshell#13>", line 1, in <module>
X.describe
AttributeError: 'numpy.ndarray' object has no attribute 'describe'
>>> X.describe()
Traceback (most recent call last):
File "<pyshell#14>", line 1, in <module>
X.describe()
AttributeError: 'numpy.ndarray' object has no attribute 'describe'
>>> X.dtype
dtype('float64')
>>> X = np.arry (df.iloc[:, column != 'netgain'])
Traceback (most recent call last):
File "<pyshell#16>", line 1, in <module>
X = np.arry (df.iloc[:, column != 'netgain'])
AttributeError: module 'numpy' has no attribute 'arry'
>>> X = np.array (df.iloc[:, column != 'netgain'])
Traceback (most recent call last):
File "<pyshell#17>", line 1, in <module>
X = np.array (df.iloc[:, column != 'netgain'])
NameError: name 'column' is not defined
>>> X = np.array (df.iloc[:, columns != 'netgain'])
Traceback (most recent call last):
File "<pyshell#18>", line 1, in <module>
X = np.array (df.iloc[:, columns != 'netgain'])
NameError: name 'columns' is not defined
>>> X = np.array(df.iloc[:,df.columns !='netgain'])
arr = randn(4,4)
np.where(arr>0,2,-2)# if arr is greater than 0, insert 2, otherwise, insert -2
# consider Excel's if function: if(boolean_testing, value_if_true, value_if_false)
# both np.where and Excel's if function can be nested, e.g.
cond1 = randn(1,100)>0
cond2 = randn(1,100)<0
result = np.where(cond1 & cond2,0, np.where(cond1,1,np.where(cond2,2,3)))
#alternatively, use boolean expression
result = 1*cond1 + 2*cond2 + 3* -(cond1 | cond2) #but more difficult to come to first thought
# statistical:
np.mean(arr,axis=0)
np.std(arr,ddof=1)#denominator: N - ddof, default 0
# boolean arrays
bools = np.arry([True,False,False,True,False])
bools.all
bools.any
# sorting
arr = randn(8)
arr.sort() # will return a copy of the sorted array
# unique
arr.unique() # applies to both numerics and strings
# save binary data
arr = np.arrange(10)
np.save('some_array.npy',arr) # save on the disk, if name not appended .npy, will append automatically
np.load('som_array.npy') # load array
np.savez('array_archive.npz',a=arr,b=arr) # save as a zip archive of arrays
#when loading an archive, the result will be a dictionary like object
ids = []
imagePaths = [os.path.join(path, f) for f in os.listdir(path)]
#print(imagePaths)
face_engine = cv.CascadeClassifier(
'/home/pi/Downloads/opencv-3.4.0/data/haarcascades/haarcascade_frontalface_default.xml'
)
for imagePath in imagePaths:
PIL_img = Image.open(imagePath).convert('L')
img_numpy = np.array(PIL_img, 'uint8')
faces = face_engine.detectMultiScale(img_numpy,
scaleFactor=1.1,
minNeighbors=5)
#print(os.path.split(imagePath))
id = int(os.path.split(imagePath)[1].split('.')[0])
#id = int(os.path.split(imagePath)[-1].split('.')[0])
for x, y, w, h in faces:
facesSamples.append(img_numpy[y:y + h, x:x + w])
ids.append(id)
return faceSamples, ids
if __name__ == '__main__':
#tu pian lu jing
path = './data/da/'
faces, ids = getImageAndLabels(path)
recognizer = cv.face.LBPHFaceRecognizer_create()
recognizer.train(faces, np.arry(ids))
recognizer.write('tr/tr.yml/')
def trnsFnc(self, entrada, zm1):
self.preintval = -arry(sgn(self.integral - entrada) * self.gain) * arry(
power(abs(self.integral - entrada), potencia(self.orden))) + zm1
def callback(self, irvalue):
frontIRvalueCaptured = 0
sideIRvalueCaptured = 0
for i in irvalue.motor_states:
if i.name == "Side_IR":
self.ir_samples[1].append(i.pulse)
#print self.ir_sample_count," side ir : ",i.pulse
sideIRvalueCaptured = 1
if i.name == "Front_IR":
self.ir_samples[0].append(i.pulse)
#print self.ir_sample_count," front ir : ",i.pulse
frontIRvalueCaptured = 1
if sideIRvalueCaptured and frontIRvalueCaptured:
self.ir_sample_count += 1
break
if not sideIRvalueCaptured and not frontIRvalueCaptured:
print "No IR sensor data received"
return
if self.ir_sample_count > 10:
self.ir_samples[1][self.ir_sample_count - 1] = np.median(
np.array(self.ir_samples[1])[-10:]) #For Front IR sensor
self.ir_samples[0][self.ir_sample_count - 1] = np.median(
np.array(self.ir_samples[0])[-10:]) #For Side IR sensor
#self.plotlist.append(self.side_ir_pulse)
frontDistance = computeDistance(
self.ir_samples[0][self.ir_sample_count - 1], True)
sideDistance = computeDistance(
self.ir_samples[1][self.ir_sample_count - 1], False)
self.CalculateDeltaPulse()
goal = pololu_trajectoryGoal()
traj = goal.joint_trajectory
traj.header.stamp = rospy.Time.now()
traj.joint_names.append(names[0])
pts = JointTrajectoryPoint()
pts.time_from_start = rospy.Duration(1.0)
if frontDistance > 300: # No wall in front
if self.turningRight: #Now there is no wall in front. So, set "turningRight" to False.
self.turningRight = False
if sideDistance == 50: # Too close to the wall
moveMotorLeft(
Max_left_limit, pts
) # stear with max left angle (should be changed with function)
elif sideDistance > 250: # Too far from the wall
moveMotorRight(
Max_right_limit, pts
) # stear with max left angle (should be changed with function)
elif sideDistance > 160:
moveMotorRight(self.side_ir_delta_pulse, pts)
elif sideDistance < 150:
moveMotorLeft(self.side_ir_delta_pulse, pts)
else:
moveMotorCenter(pts)
else:
if frontDistance < 200:
if self.turningRight:
moveMotorRight(Max_right_limit, pts)
else:
moveMotorLeft(Max_left_limit, pts)
elif sideDistance == 50: # Too close to the wall
moveMotorLeft(
Max_left_limit, pts
) # stear with max left angle (should be changed with function)
elif sideDistance > 300: #If no wall is detected by right side sensor, then take right turn
self.turningRight = True
moveMotorRight(Max_right_limit, pts)
elif sideDistance > 160:
moveMotorRight(self.side_ir_delta_pulse, pts)
elif sideDistance < 150:
moveMotorLeft(self.side_ir_delta_pulse, pts)
else:
moveMotorCenter(pts)
pts.velocities.append(1.0)
traj.points.append(pts)
client.send_goal(goal)
#if self.ir_sample_count == 1000:
# print len(self.plotlist)
# fig, ax =plt.subplots()
# plt.subplot(1,2,1),plt.plot(range(1,self.ir_sample_count+1),self.ir_samples[1],label="raw")
# plt.subplot(1,2,2),plt.plot(range(1,len(self.plotlist)+1),self.plotlist,label="median")
# ax.set_axisbelow(True)
# ax.set_ylim(80,200)
# plt.legend(loc="center right")
# plt.show()
if ir_sample_count > 100:
temp = np.arry(self.ir_samples[1])[-11:]
del ir_samples[1][:]
self.ir_samples[1] = temp
temp = np.arry(self.ir_samples[0])[-11:]
del ir_samples[0][:]
self.ir_samples[0] = temp
self.ir_sample_count = 10
client.wait_for_result(rospy.Duration.from_sec(0.005))
import caffe
import numpy as np
import sys
if len(sys.argv) != 3:
print "Usage: python convert_protomean.py proto.mean out.npy"
sys.exit()
blob = caffe.proto.caffe_pb2.BlobProto()
data = open(sys.argv[1], 'rb').read()
blob.ParseFromString(data)
arr = np.arry(caffe.io.blobproto_to_array(blob))
out = arr[0]
np.save(sys.argv[2], out)
NameError: name 'arr' is not defined
import numpy as np
lst=[1,2,3]
for i in lst:
print(i**2)
arr = np.array(lst)
arr
arr**2
arr = np.arry([1,'two',3])
arr
1
4
9
Traceback (most recent call last):
File "<ipython-input-8-b3566ca72dc1>", line 12, in <module>
arr = np.arry([1,'two',3])
File "C:\Users\user\anaconda3\lib\site-packages\numpy\__init__.py", line 220, in __getattr__
"{!r}".format(__name__, attr))
AttributeError: module 'numpy' has no attribute 'arry'
def moving_average(vals, n):
out = []
for i in range(len(val) - n - 1):
out[i] = vals[i:i + n].mean()
return np.arry(vals)
def trnsFnc(self,entrada,zm1): self.preintval=-arry(sgn(self.integral-entrada)*self.gain)*arry(power(abs(self.integral-entrada),potencia(self.orden)))+zm1
# -*- coding: utf-8 -*- """ Created on Tue Jun 30 08:45:53 2015 @author: ABerner """ # Code snippet from pg 200 import numpy as np from quantecon import mc_sample_path P = np.array([[.4,.6],[.2,.8]]) s = mc_sample_path(P, init=(0.5, 0.5), sample_size=100000) print((s ==0).mean()) # Example: Powers of a Markov Matrix on page 201 P = np.array([[.971,.029,0],[.145,.778,0.077],[0,0.508,0.492]]) psi = np.arry([0.80,0.19,0.01]) #hypothetical state probability estimate vector np.inner(np.dot(psi,np.linalg.matrix_power(P,6)),np.array([0,1,1])) #6m forward forecast of recession probability given the current state estimates
