iris >> dp.mutate(redondeado=X.PetalLength.round(), redondeado2=X.SepalLength.round())
iris.assign(redondeado = lambda x: x.PetalLength.round(), redondeado2 = lambda x: x.SepalLength.round())
#ifelse(y==0, 0, 1)
np.where((y == 0), 0, 1)
# data=iris %>%
# distinct(Species, Sepal.Length, .keep_all = T)
iris >> dp.distinct(X.SepalLength)
iris.drop_duplicates()
iris.drop_duplicates(subset='PetalLength')
# #ordenando
# data=iris %>%
# arrange(Sepal.Length, Sepal.Width)
iris >> dp.arrange(X.PetalLength)
iris.sort_values("PetalLength", ascending=False)
# data$ceros=0
iris['ceros']=0
# data$ceros=NULL
del(iris['ceros'])
# data=data.frame(x1=rep(1,10), x2=rep(2, 10))
data = pd.DataFrame({'x1':np.ones(10), 'x2':np.repeat(2, 10)})
# data2=data.frame(x3=rep(3,10), x4=rep(4, 10))
data2 = pd.DataFrame({'x3':np.repeat(3, 10), 'x4':np.repeat(4, 10)})
# data3=data.frame(x1=rep(1,10), x5=rep(5, 10))
data3 = pd.DataFrame({'x1':np.ones(10), 'x5':np.repeat(5, 10)})
# total=left_join(data, data3, by="x1")
# total2=merge(data, data2, by="x1")
def main(argv):
yURL = None
outdir = None
maxFrames = 500
yURL = input("Enter the youtube url:")
outdir = input("Enter the output directory:")
maxFrames = int(input("Enter the maximum number of frames to check:"))
faceDet = cv2.CascadeClassifier(
"haarcascade/haarcascade_frontalface_default.xml")
faceDet2 = cv2.CascadeClassifier(
"haarcascade/haarcascade_frontalface_alt2.xml")
faceDet3 = cv2.CascadeClassifier(
"haarcascade/haarcascade_frontalface_alt.xml")
faceDet4 = cv2.CascadeClassifier(
"haarcascade/haarcascade_frontalface_alt_tree.xml")
#
pdata, pframes, pfacedims = getNewInstances(yURL,
faceDet,
faceDet2,
faceDet3,
faceDet4,
maxCount=maxFrames)
#
headers = dict()
headers['Ocp-Apim-Subscription-Key'] = ms_key1
headers['Content-Type'] = 'application/octet-stream'
#
resultsDf = pd.DataFrame()
frameId = 0
for image in pframes:
print("posting frame %d of %d" % (frameId, len(pframes)))
#sending the frame image to MS cognitive services
resultMS = processRequest(image, headers)
#isinstance == type()
if isinstance(resultMS, list):
for result in resultMS:
if isinstance(result, dict):
resFrameList = []
for res in result['scores'].items():
resFrameList.append(
(frameId, res[0], res[1],
result["faceRectangle"]['left'],
result["faceRectangle"]['top'],
result["faceRectangle"]['width'],
result["faceRectangle"]['height']))
appendDf = pd.DataFrame(resFrameList,
columns=[
"frameId", "emotionLabel",
"conf", "faceleft",
"facetop", "faceW", "faceH"
])
resultsDf = resultsDf.append(appendDf)
time.sleep(2)
frameId += 1
#
# print(resultsDf)
#we append all the data to the dataframe
#http://bluescreen.club/2017/06/18/import-pandas-as-pd/
#then we convert the dataframe to a Dplyframe object which allows us to do higher level data analytics
#for this one, we will select out the top most ranking face frames for each of the emotions
#microsoft provides us with around 8 emotions
#so we sort out 8 faces for 8 emotions and then save them accordingly
dfFaces = DplyFrame(resultsDf)
# print(dfFaces)
topFaces = (
dfFaces >> group_by(X.emotionLabel) >> sift(X.conf == X.conf.max()) >>
sift(X.frameId == X.frameId.min()) >> ungroup() >> group_by(
X.frameId) >> sift(X.conf == X.conf.max()) >> ungroup() >> arrange(
X.emotionLabel))
topFaces = topFaces.drop_duplicates()
#print(topFaces)
i = 0
for index, row in topFaces.iterrows():
print("saving emotion frame %d of %d" % (i, len(topFaces.index)))
#
emotion = row["emotionLabel"]
confid = int(row["conf"] * 100)
image = pframes[int(row["frameId"])]
faceL = row["faceleft"]
faceT = row["facetop"]
faceW = row["faceW"]
faceH = row["faceH"]
#save cropped face
imageW = image[faceT:faceT + faceH, faceL:faceL + faceW]
cv2.imwrite(
os.path.expanduser("%s/Cropped_%s.jpg" % (outdir, emotion)),
imageW)
#if you wish to put a rectangle on the faces then uncomment below
#
# cv2.rectangle( image,(faceL,faceT),
# (faceL+faceW, faceT + faceH),
# color = (255,0,0), thickness = 5 )
# cv2.putText( image, emotion, (faceL,faceT-10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255,0,0), 1 )
#
cv2.imwrite(os.path.expanduser("%s/%s.jpg" % (outdir, emotion)), image)
i += 1
"""5.2 wordcloud"""
"""turns the word freq to dict"""
d = {}
for a, x in dff.values:
d[a] = x
wordcloud = WordCloud(width = 1000, height = 1000,
background_color ='white',
min_font_size =15, max_font_size=120).generate_from_frequencies(frequencies=d)
plt.figure(figsize = (8, 8), facecolor = None)
plt.imshow(wordcloud)
plt.axis("off")
plt.tight_layout(pad = 0)
plt.show()
"""stacked bar plot"""
dfx = (dfr >>
arrange(-X.tot))
dfx=dfx.head(50)
from plotnine import *
(ggplot(dfx, aes(x='word', y='tot', fill='source'))
+ geom_col() +
theme(axis_text_x=element_text(rotation=45, hjust=1))
)
"""each newspaper"""
dfr = DplyFrame(output)
df_tele =(dfr >>sift(X.source=="guardian"))
df_tele = (df_tele >>
group_by(X.word, X.source) >>
summarize(tot=X.count.sum()))
df_tele = (df_tele >>select(X.word, X.tot ))
>> pull('carat'))
# DataFrame transformation
diamonds >> mutate(x_plus_y=X.x + X.y) >> select(columns_from('x')) >> head(3)
diamonds >> mutate(x_plus_y=X.x + X.y, y_div_z=(X.y / X.z)) >> select(columns_from('x')) >> head(3)
# The transmute() function is a combination of a mutate and a selection of the created variables.
diamonds >> transmute(x_plus_y=X.x + X.y, y_div_z=(X.y / X.z)) >> head(3)
# group_by() and ungroup()
diamonds >> head(5) >> group_by(X.color) >> mutate(avg_price=X.price.mean())
(diamonds >> group_by(X.cut) >>
mutate(price_lead=lead(X.price), price_lag=lag(X.price)) >>
head(2) >> select(X.cut, X.price, X.price_lead, X.price_lag))
# ungroup()
(diamonds >> group_by(X.cut) >> arrange(X.price) >>
head(3) >> ungroup() >> mask(X.carat < 0.23))
# Reshaping
diamonds >> arrange(X.table, ascending=False) >> head(5)
diamonds >> rename(CUT=X.cut, COLOR='color') >> head(2)
# transform a "wide" DataFrame into a "long" format
# gather(key, value, *columns, add_id=True) melts the specified columns into two key:value columns
diamonds >> gather('variable', 'value', ['price', 'depth','x','y','z']) >> head(5)
elongated = diamonds >> gather('variable', 'value', add_id=True)
# transform a "long" DataFrame into a "wide" format
# In this case the _ID column comes in handy since it is necessary to not have any duplicated identifiers.
widened = elongated >> spread(X.variable, X.value)
widened.dtypes # all are objects
def main(argv):
ytURL = None
outdir = None
maxFrames = 500
try:
opts, args = getopt.getopt(argv, "hy:o:m:",
["yturl=", "odir=", "maxframes="])
except getopt.GetoptError:
print 'Error: shellScript.py -y <yturl> -o <odir> -m <maxframes>'
sys.exit(2)
#print opts
for opt, arg in opts:
if opt == '-h':
print 'help: shellScript.py -y <yturl> -o <odir> -m <maxframes>'
sys.exit()
elif opt in ("-y", "--yturl"):
print("--yturl={}".format(arg))
ytURL = arg
elif opt in ("-o", "--odir"):
print("--odir={}".format(arg))
outdir = arg
elif opt in ("-m", "--maxframes"):
print("--maxframes={}".format(arg))
maxFrames = int(arg)
#
if ytURL is None:
print 'bad yt: shellScript.py -y <yturl> -o <odir> -m <maxframes>'
sys.exit()
#
if outdir is None:
print 'bad outdir: shellScript.py -y <yturl> -o <odir> -m <maxframes>'
sys.exit()
#
if False == isinstance(maxFrames, (int, long)):
print 'bad maxFrames: shellScript.py -y <yturl> -o <odir> -m <maxframes>'
sys.exit()
#
#
faceDet = cv2.CascadeClassifier(
"haarcascade/haarcascade_frontalface_default.xml")
faceDet2 = cv2.CascadeClassifier(
"haarcascade/haarcascade_frontalface_alt2.xml")
faceDet3 = cv2.CascadeClassifier(
"haarcascade/haarcascade_frontalface_alt.xml")
faceDet4 = cv2.CascadeClassifier(
"haarcascade/haarcascade_frontalface_alt_tree.xml")
#
pdata, pframes, pfacedims = getNewInstances(ytURL,
faceDet,
faceDet2,
faceDet3,
faceDet4,
maxCount=maxFrames)
#
headers = dict()
headers['Ocp-Apim-Subscription-Key'] = ms_key1
headers['Content-Type'] = 'application/octet-stream'
#
resultsDf = pd.DataFrame()
frameId = 0
for image in pframes:
print("posting frame %d of %d" % (frameId, len(pframes)))
resultMS = processRequest(image, headers)
#
if isinstance(resultMS, list):
for result in resultMS:
if isinstance(result, dict):
resFrameList = []
for res in result['scores'].items():
resFrameList.append(
(frameId, res[0], res[1],
result["faceRectangle"]['left'],
result["faceRectangle"]['top'],
result["faceRectangle"]['width'],
result["faceRectangle"]['height']))
appendDf = pd.DataFrame(resFrameList,
columns=[
"frameId", "emotionLabel",
"conf", "faceleft",
"facetop", "faceW", "faceH"
])
resultsDf = resultsDf.append(appendDf)
time.sleep(2)
frameId += 1
#
dfFaces = DplyFrame(resultsDf)
#
topFaces = (
dfFaces >> group_by(X.emotionLabel) >> sift(X.conf == X.conf.max()) >>
sift(X.frameId == X.frameId.min()) >> ungroup() >> group_by(
X.frameId) >> sift(X.conf == X.conf.max()) >> ungroup() >> arrange(
X.emotionLabel))
topFaces = topFaces.drop_duplicates()
#print(topFaces)
#
i = 0
for index, row in topFaces.iterrows():
print("saving emotion frame %d of %d" % (i, len(topFaces.index)))
#
emotion = row["emotionLabel"]
confid = int(row["conf"] * 100)
image = pframes[int(row["frameId"])]
faceL = row["faceleft"]
faceT = row["facetop"]
faceW = row["faceW"]
faceH = row["faceH"]
#
#save cropped face
imageW = image[faceT:faceT + faceH, faceL:faceL + faceW]
cv2.imwrite(
os.path.expanduser("%s/Cropped_%s.jpg" % (outdir, emotion)),
imageW)
#
cv2.rectangle(image, (faceL, faceT), (faceL + faceW, faceT + faceH),
color=(255, 0, 0),
thickness=5)
cv2.putText(image, emotion, (faceL, faceT - 10),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 0, 0), 1)
#
cv2.imwrite(os.path.expanduser("%s/box%s.jpg" % (outdir, emotion)),
image)
i += 1