def confirm(self, controller):
# se si tratta di Verifica del paziente
if self.role == 0:
print("Verifica del paziente")
patient, index, user = Recognition.recognize(
self.cf.cget("text")[16:], self.photo)
# se si tratta di Verifica del delegato
else:
print("Verifica del delegato")
patient, index, user = Recognition.identify(
self.cf.cget("text")[16:], self.photo)
print("Il paziente e':", patient, "L'utente riconosciuto e':", user)
if user is not None:
list(controller.frames.values())[pages.UserPage.value - 1].reset()
list(controller.frames.values())[pages.UserPage.value -
1].update_data(
index, user["User"],
patient["User"],
patient["Codice Fiscale"],
patient["Delegati"],
patient["Farmaci"],
patient["Data"],
self.panel.image)
controller.show_frame(UserPage)
else:
list(
controller.frames.values())[pages.InformationPage.value -
1].update_data(recognitionRejected)
controller.show_frame(InformationPage)
def xzdingwei():
lable3 = tk.Label(window, text="形状定位如下", font=('Arial', 15))
lable3.place(x=750, y=0)
Recognition.find_car_num_brod(pic_path)
img = Image.open("tmp/chepai_img1.jpg")
photo = ImageTk.PhotoImage(img)
img = img.resize((160, 40), Image.ANTIALIAS)
photo = ImageTk.PhotoImage(img)
label2 = tk.Label(window,anchor="ne")
label2.config(image=photo)
label2.image=photo
label2.place(x=735,y=30)
def saveFrame(self, userId):
""" write frame to disk and add row to file config
Return:
detection - face is detected
newUser - created new user
"""
imgBasePath = ""
imgBasePath = self.faceDir + "/" + "face_" + userId
recognition = rec.Recognition(False)
# faceImg, detection, x, y, h, w = recognition.detectFace(
# cv2.cvtColor(self.currentFrame, cv2.COLOR_RGB2GRAY))
# faceImg2, detection2 = recognition.getCroppedImageByEyes(
# cv2.cvtColor(self.currentFrame, cv2.COLOR_RGB2GRAY), (0.2, 0.2))
#faceImg, detection = recognition.getCroppedImageByEyes(
faceImg, detection, x, y, h, w = recognition.detectFace(
cv2.cvtColor(self.currentFrame, cv2.COLOR_RGB2GRAY))
if detection:
writer = rwi.ReadWriteIndex()
faceImg = cv2.resize(faceImg, (92, 112))
fileExtension = "." + self.config.get("Cam", "imgExtension")
filePath = (imgBasePath + "#" +
str(writer.getCountUserElem(userId)) + fileExtension)
cv2.imwrite(self.currentDir + filePath, faceImg)
newUser = writer.checkUser(userId)
writer.addRow(userId, filePath) # add new line to file index
return detection, not newUser
return detection, 0
def updateThreshold(new_user):
gallery_threshold = np.load("npy_db/gallery_thresholds.npy").tolist()
gallery_target = np.load("npy_db/gallery_target.npy")
gallery_histograms = np.load("npy_db/histogram_gallery_data.npy")
new_index = gallery_target.tolist().index(new_user)
max = -1
# ritorna la lista di utenti unici. Si usa il dizionario perche' cosi' l'ordine tra gallery users e gallery
# thresholds e' lo stesso
galley_users = list(dict.fromkeys(gallery_target))
print("AGGIORNAMENTO DEI THRESHOLDS...")
for user in galley_users:
# confrontiamo un utente con tutti gli utenti della gallery ma non con se stesso
if user != new_user:
index = galley_users.index(user)
# per ogni template dell'utente
for i in range(n_photo_x_user):
thd = Recognition.topMatch(user, gallery_target,
gallery_histograms,
gallery_histograms[new_index + i])
# prendo il threshold massimo e lo arrotondo
if thd > gallery_threshold[index]:
gallery_threshold[index] = thd
if thd > max:
if np.round(thd, 2) <= thd:
max = np.round(thd, 2) + 0.01
else:
max = np.round(thd, 2)
# l'ultimo max fa riferimento all'ultimo utente appena iscritto, questo proprio perche' si segue l'ordine di
# gallery_users, che mantiene a sua volta l'ordine di gallery_target
gallery_threshold.append(max)
print("IL TUO THRESHOLD:", max, "N. TOTALI DI HISTOGRAM:",
len(gallery_threshold))
np.save("npy_db/gallery_thresholds.npy", gallery_threshold)
return
def __init__(self, video_src=None, modelType=None, threshold=None):
QtGui.QMainWindow.__init__(self)
self.recognition = rec.Recognition(True, modelType, threshold)
self.setWindowTitle("Authorization system - Face")
cWidget = QtGui.QWidget(self)
mainLayout = QtGui.QVBoxLayout()
# Title
# titleLabel = QtGui.QLabel("Show your credential!")
# titleLabel.setAlignment(QtCore.Qt.AlignCenter)
# Webcam
self.imgLabel = QtGui.QLabel()
self.webcamSampling = vs.VideoSampling(video_src)
self.update()
# Button
self.startButton = QtGui.QPushButton("Authenticate")
self.cancelButton = QtGui.QPushButton("Cancel")
self.connect(self.startButton, QtCore.SIGNAL("clicked()"),
self.startAuthentication)
self.connect(self.cancelButton, QtCore.SIGNAL("clicked()"),
QtCore.SLOT("close()"))
# mainLayout.addWidget(titleLabel)
mainLayout.addWidget(self.imgLabel)
mainLayout.addWidget(self.startButton)
mainLayout.addWidget(self.cancelButton)
mainLayout.setAlignment(QtCore.Qt.AlignCenter)
cWidget.setLayout(mainLayout)
self.setCentralWidget(cWidget)
self.center()
def adaptiveThresholds(self):
print("ADAPTIVE THRESHOLDS")
thresholds = []
# lista di utenti non ripetuti
galley_users = list(dict.fromkeys(self.gallery_target))
# per ogni utente
for user in galley_users:
max_thd = -1
# per ogni template
for i in range(len(self.gallery_data)):
# se i due utenti non sono uguali
if user != self.gallery_target[i]:
new_thd = 0
# ottengo l'histogram h del template
hist_probe = self.histogram_gallery_data[i]
index = self.gallery_target.index(user)
# per ogni histogram dell'utente, lo comparo con h e prendo il valore di similarity piu' alto
for j in range(self.get_n_gallery_temp_x_user()):
hist_gallley = self.histogram_gallery_data[index + j]
diff = Recognition.compareHistogram(
hist_probe, hist_gallley)
if diff >= new_thd:
new_thd = diff
# aggiorno il threshold piu' alto ottenuto per user e lo arrotondo
if new_thd > max_thd:
if np.round(new_thd, 2) <= new_thd:
max_thd = np.round(new_thd, 2) + 0.01
else:
max_thd = np.round(new_thd, 2)
thresholds.append(max_thd)
print("Threshold per l'utente", user, ":", max_thd)
print("Thresholds:", thresholds)
return thresholds
def SendMouseClick(a):
x, y = Recognition.getXYCoord(a)
print(x,y)
if (x != -999 & y != -999):
Cursor.MoveCursor(x,y)
pyautogui.click(x,y)
time.sleep(0.5)
def convert(*args):
# call extraction functions on the image, return output
im = PIL.Image.open(imPath.get())
extracts = extraction3.main3(im)
net = Recognition.load('network_improved4.json')
output_string = net.recognize(extracts)
t.set(output_string)
ttk.Label(mainframe, textvariable=t).grid(column=4, row=3)
def update_btn_text2():
butt1.config(state=NORMAL)
v1.set("Your system is armed")
print(butt1['state'])
print("rec wcalled")
#while butt1['state']== 'normal':
print("entered loop")
x = Recognition.eigrec(names)
print(x)
em2.eml(x)
print("lopoexit")
butt2.config(state=DISABLED)
def recognition(queue_images, queue_number):
"""
This gets the images stored in the queue and use the recognition algorithm established in the file "Recognition".
Then it put the number in an other queue.
:param queue_images: the queue containing the images detected
:param queue_number: the queue containing the number recognized
"""
# Global loop
while True:
print("Queue size :", queue_images.qsize())
images, type = queue_images.get()
if images is not None:
for image in images:
number = Recognition.detect_number(image)
queue_number.put((number, type))
imgTrafficLabel = tkinter.Label(canvasTraffic, image=imgTraffic)
windowSpeedLimit = tkinter.Tk()
windowSpeedLimit.title("Speed limitation")
canvasSpeedLimit = tkinter.Canvas(windowSpeedLimit)
img = ImageTk.PhotoImage(file="assets/blank.jpg")
sign = tkinter.Label(canvasSpeedLimit, image=img)
text = tkinter.Label(windowSpeedLimit, text="")
for elt in Images:
imgs = Processing.pre_processing(elt)
print("processed")
img = ImageTk.PhotoImage(file="assets/blank.jpg")
if imgs is not None:
for image in imgs:
number = Recognition.detect_number(image)
if number is not None:
if "11" in number or "10" in number:
print("@@@@ 110 @@@@")
img = ImageTk.PhotoImage(file="assets/ref110.jpg")
elif "30" in number:
if "130" in number or "13" in number:
print("@@@@ 130 @@@@")
img = ImageTk.PhotoImage(file="assets/ref130.jpg")
else:
print("@@@@ 30 @@@@")
img = ImageTk.PhotoImage(file="assets/ref30.jpg")
elif "5" in number:
print("@@@@ 50 @@@@")
img = ImageTk.PhotoImage(file="assets/ref50.jpg")
elif "7" in number:
import keyboard
from Frame import Framework
from ChessInformation import chessInformation
import json
from Recognition import *
picutre_path = "D:/MyProject/python/testLOL/picture/"
frame = Framework()
imgHandle = chessInformation()
cnt = 0
with open(r'D:\MyProject\python\testLOL\project\scr\information.json',
'r') as f:
config = json.load(f)
rec = Recognition(config)
def Id2Name(id):
if id in config:
return config[id]["HeroName"]
else:
return "None"
while True:
# print("get " + str(cnt))
keyboard.wait("f9")
for i in range(5):
img = imgHandle.getChessImg(i, frame)
print(Id2Name(rec.get(img)))
# time.sleep(20)
USAGE
python main.py -i videos/sj.mp4 -d dataset
"""
import Recognition
import Encoder
import argparse
# read arguments
ap = argparse.ArgumentParser()
ap.add_argument("-d",
"--dataset",
type=str,
default="dataset",
help="input face dataset path")
ap.add_argument("-i", "--input", type=str, help="path to input video")
args = vars(ap.parse_args())
# start encoding
print("Encoding data set...")
encoder = Encoder.Encoder(args['dataset'])
encode_file_dir = encoder.encode()
print("Completely encode the data set")
# start detection and recognition
print("Detecting the face...")
input_video_dir = args['input']
recognition = Recognition.Recognition(encode_file_dir, input_video_dir)
mosaic_list = recognition.recognize()
print("Completely detected faces from input video")
def parse(str):
#start up the emulator
if (str == "-init"):
Start.start()
Done = TerminateKeyInput("icon.png", "enter", "App opened.")
if (Done != ""): return Done
else: return "Error starting up the app."
#return to home
if (str == "-h"):
Start.openApp()
SendKeyInput("home.png", "h")
#run dimension missions
if (str == "-dm"):
Start.openApp()
SendMouseClick("home.png")
SendMouseClick("enter.png")
SendMouseClick("dm.png")
SendMouseClick("ready.png")
SendMouseClick("clear.png")
SendMouseClick("cmax.png")
SendMouseClick("home.png")
noEnergyErr = TerminateKeyInput("noenergy.png", "n", "Out of energy. Returning to home.")
if noEnergyErr != "":
SendKeyInput("home.png", "h")
return noEnergyErr
time.sleep(4)
SendMouseClick("close.png")
Done = TerminateKeyInput("home.png", "h", "Done.")
if Done != "":
return "Done."
else: return "Error."
#collect antimatter from store
if (str == "-lab col"):
Start.openApp()
SendMouseClick("home.png")
SendMouseClick("list.png")
SendMouseClick("lab.png")
SendMouseClick("amcollect.png")
time.sleep(0.5)
Done = TerminateKeyInput("home.png", "h", "Antimatter collected. Returning to home.")
if Done != "":
return Done
else:
return "An error occurred."
#Legendary battle: Loki Ragnarok
if (str == "-lb"):
Start.openApp()
SendMouseClick("start.png")
pyautogui.press("l")
time.sleep(10)
pyautogui.press("l")
while True:
if (Recognition.locate("repeat.png")):
pyautogui.press("l")
time.sleep(4)
return "Stage complete."
pyautogui.press("l")
pyautogui.press("l")
time.sleep(0.5)
pyautogui.press(";")
pyautogui.press(";")
time.sleep(0.5)
pyautogui.press("'")
pyautogui.press("'")
time.sleep(0.5)
pyautogui.press("[")
pyautogui.press("[")
time.sleep(0.5)
pyautogui.press("p")
pyautogui.press("p")
time.sleep(0.5)
return ""
def TerminateKeyInput(a, b, c):
if (Recognition.locate(a)) == 1:
pyautogui.press(b)
return c
else:
return ""
def SendKeyInput(a, b):
if (Recognition.locate(a)) == 1:
pyautogui.press(b)
#model.plotResults() """print "body points" print gmB1.b_points print "gravity points" print gmB1.gr_points print "body sigma" print gmB1.b_sigma print "gravity sigma" print gmB1.gr_points""" from Recognition import* r = Recognition(list_models) #Calculate best weights print "Calculating weigths" r.calculate_Weights() print "Validation" sfile = "validation/t1.txt" r.recognition_from_files(sfile) sfile = "validation/t2.txt" r.recognition_from_files(sfile)
import Calibration
import Recognition
import IOFunctions
import Sequence
import math
MIN_SEQ_LENGTH = 100
MAX_SEQ_LENGTH = 150
def initial():
dataL = IOFunctions.read('Data\\test2_B.csv')
dataR = IOFunctions.read('Data\\test3_B.csv')
#return {"Rechts" : [Sequence.Sequence(dataR, 1200, 100).getNormalized(), Sequence.Sequence(dataR, 1982, 100).getNormalized()],
# "Links" : [Sequence.Sequence(dataL, 2506, 100).getNormalized(), Sequence.Sequence(dataL, 2672, 100).getNormalized()]}
return {"Rechts" : [Sequence.Sequence(dataL, 2672, 100).getNormalized(), Sequence.Sequence(dataL, 2506, 100).getNormalized()],
"Links" : [Sequence.Sequence(dataL, 2506, 100).getNormalized(), Sequence.Sequence(dataL, 2672, 100).getNormalized()]}
def getAllLengths():
allLengths = []
for seqLength in range(MIN_SEQ_LENGTH,MAX_SEQ_LENGTH+1,math.ceil(MAX_SEQ_LENGTH*0.03)):
allLengths.append(seqLength)
return allLengths
if __name__ == '__main__':
allLengths = getAllLengths()
labels = Calibration.run(["Rechts","Links"], allLengths)
#labels = initial()
Recognition.recognize(labels, MIN_SEQ_LENGTH, allLengths)
#!/usr/bin/env python # coding: utf-8 # In[3]: import mnist_loader training_data, validation_data, test_data = mnist_loader.load_data_wrapper() training_data = list(training_data) # In[2]: import Recognition # In[6]: net = Recognition.Network([784, 100, 10]) net.SGD(training_data, 30, 10, 0.001, test_data=test_data) # In[3]: net = Recognition.Network([784, 30, 10]) # In[4]: net.SGD(training_data, 30, 10, 3.0, test_data=test_data) # In[4]: import network2 # In[5]:
# script for making a saving a neural network
import Recognition
import Data
# get the data
train_data, test_data = Data.alpha_load()
net = Recognition.load('network_improved3.json')
net.train(train_data, 5, 104, 0.1, reg_param=5.0, test_data=test_data)
net.save('network_improved4.json')
print "Network Saved"
def read(self):
print("hi")
print(Recognition.readNumber(self.getter()))
