def test_funcs(img1, img2):
img3 = img2
funcs.show(img1)
funcs.save(img1, filename='out_1.png')
#funcs.getWidth funcs.getHeight
print funcs.getWidth(img1), funcs.getHeight(img1)
#funcs.mix
mixed_pics = funcs.mix(img1, img3, .5)
funcs.show(mixed_pics)
#funcs.tint
tinted_pic = funcs.tint(img1, (0, 0, 255), .5)
funcs.show(tinted_pic)
#funcs.grayscale
gray_pic = funcs.grayscale(img1)
funcs.show(gray_pic)
gray_pic_2 = funcs.grayscale(img1, two_dimensional=True)
funcs.show(gray_pic_2)
#funcs.rotate
img_rotate = funcs.rotate(img1, 180)
funcs.show(img_rotate)
def write_temperatures(temperatures):
date = datetime.datetime.now().strftime('%Y%m%d')
try:
df = load('history/' + date + '.df')
df.loc[df.index.max() + 1] = temperatures
except FileNotFoundError:
df = pd.DataFrame(temperatures, index=[0])
save(df, 'history/' + date + '.df')
def getSeam(img, blur=1, debug=False, attractor=np.zeroes(img.shape)):
img_original = img.copy()
img = img.copy()
img = funcs.grayscale(img, True)
img = np.float32(cv2.GaussianBlur(img, (blur, blur), 0))
derivative_kernel = np.float32([[-1, 0, 1], [-2, 0, 2], [-1, 0, 1]])
Ix = cv2.filter2D(img, -1, derivative_kernel)
Iy = cv2.filter2D(img, -1, derivative_kernel.T)
path = Ix**2 + Iy**2
path = path**.5
path += (atrractor * 100.0)
if debug:
pic_1_c_0 = path.copy()
funcs.save(pic_1_c_0, "output/pic_1_a.png")
kernel = np.float32([[1, 1, 1]])
h, w = img.shape[:2]
for i in range(1, h):
row = path[i - 1, :]
row = cv2.erode(row, kernel.T)
path[i:i + 1, :] += row.T
if debug:
pic_1_c_1 = path.copy()
funcs.save(pic_1_c_1, "output/pic_1_b.png")
x = np.argmin(path[-1])
y = h - 1
seam = []
seam.append(x)
while y > 0:
y -= 1
minX = seam[0] - 1
maxX = seam[0] + 2
x = np.argmin(path[y, max(minX, 0):min(maxX, w - 1)]) + x
if x != 0:
x -= 1
seam.insert(0, x)
if debug:
funcs.save(showSeam(pic_1_c_0, seam), "output/pic_1_c_1.png")
funcs.save(showSeam(pic_1_c_1, seam), "output/pic_1_c_2.png")
funcs.save(showSeam(img_original, seam), "output/pic_1_c_0.png")
return seam
if (r[i].loaded and not r[i].excluded):
r[i].make_groups(params, window)
msg = 'Status: %s: %d groups generated' % (r[i].title,
len(r[i].glist))
print(msg)
stext.Update(value=msg)
if (r[index].detected):
current_img = funcs.draw_circles(r[index].data, r[index].holes,
params)
if (r[index].grouped):
current_img = funcs.draw_groups(current_img, r[index].glist,
r[index].gcrd, r[index].holes,
params)
funcs.draw_image(graph, current_img, params)
elif event == 'Save' and opened:
funcs.save(r, nv, fname_out, params)
stext.Update(value='Status: Wrote centers in %s' % fname_out)
saved = True
elif event == '-GRAPH-' and mode == 't' and opened and r[index].loaded:
if mouse_x0 < 0 and mouse_y0 < 0:
mouse_x0, mouse_y0 = values['-GRAPH-']
else:
mouse_x1, mouse_y1 = values['-GRAPH-']
dx = mouse_x1 - mouse_x0
dy = mouse_y1 - mouse_y0
params.translate(dx, dy)
funcs.draw_image(graph, current_img, params)
mouse_x0 = mouse_x1
mouse_y0 = mouse_y1
elif event == '-GRAPH-' and mode == 'z' and opened and r[index].loaded:
if mouse_x0 < 0 and mouse_y0 < 0:
img = funcs.trans_img(trans_img)
seam_map = funcs.trans_img(trans_map)
return img, seam_map
#Read Image
img1 = cv2.imread("image1.png")
#pic_1_a,pic_1_b,pic_1_c_0,pic_1_c_1,pic_1_c_2
seam1 = getSeam(img1, 1, debug=True)
#pic_1_d
pic_1_d = removeSeam(img1, seam1)
funcs.save(pic_1_d, "output/pic_1_d.png")
#pic_1_e
pic_1_e, _ = carveSeams(img1, 120)
funcs.save(pic_1_e, "output/pic_1_e.png")
#pic_2_a
trans_img = funcs.trans_img(img1)
pic_2_a = showSeam(trans_img, getSeam(trans_img))
pic_2_a = funcs.trans_img(pic_2_a)
funcs.save(pic_2_a, "output/pic_2_a.png")
#pic_2_b
pic_2_b, _ = carveHorizSeams(img1, 120)
funcs.save(pic_2_b, "output/pic_2_b.png")
def run_model(init, output, saver, IS_RESTORE_BASED, company_str, company,
epochs, seq_input, seq_output, keep_prob, dropout, optimizer,
cost, batch_size, prediction_length, sequence_length,
processing_device, display_steps, weights, biases, total_error,
unexplained_error, R_squared, R, MAPE, RMSE):
merged = tf.summary.merge_all()
with tf.device(processing_device):
with tf.Session() as sess:
if IS_RESTORE_BASED:
saver.restore(
sess, "../modeldata/" + company_str + "/logs/model.ckpt")
sess.run(init)
train_writer = tf.summary.FileWriter(
'../modeldata/' + company_str + '/logs/', sess.graph)
train_start = time.time()
step = 1
while company.train.epochs_completed <= epochs:
step += 1
company_data, company_labels = company.train.next_batch()
output_data = np.reshape(company_labels.T[1].T,
(batch_size, prediction_length))
_, loss, rsq, r, te, une, mape, rmse = sess.run(
[
optimizer, cost, R_squared, R, total_error,
unexplained_error, MAPE, RMSE
],
feed_dict={
seq_input: company_data.T[1].T,
seq_output: output_data,
keep_prob: dropout
})
if step % display_steps == 0:
print "Epochs completed: {}".format(company.train.epochs_completed) +\
" loss: {}".format(loss) + " step: {}".format(step)
if company.train.epochs_completed == epochs - 1:
p_valfile = open(
"../graphdata/" + company_str + "_pval.js", "w")
p_valfile.write("var " + company_str + "_total_error = " +
str(te) + ";")
p_valfile.write("var " + company_str +
"_unexplained_error = " + str(une) + ";")
p_valfile.write("var " + company_str + "_R_squared = " +
str(rsq) + ";")
p_valfile.write("var " + company_str + "_R = " + str(r) +
";")
p_valfile.write("var " + company_str + "_MAPE = " +
str(mape) + ";")
p_valfile.write("var " + company_str + "_RMSE = " +
str(rmse) + ";")
p_valfile.close()
print "Optimization Completed. Training time: {}sec".format(
time.time() - train_start)
# save weights, biases, model:
save(sess.run(weights['layer1']),
"../modeldata/" + company_str + "/weights/layer1")
save(sess.run(weights['layer2']),
"../modeldata/" + company_str + "/weights/layer2")
save(sess.run(biases['layer1']),
"../modeldata/" + company_str + "/biases/layer1")
save(sess.run(biases['layer2']),
"../modeldata/" + company_str + "/biases/layer2")
save_path = saver.save(
sess, "../modeldata/" + company_str + "/logs/model.ckpt")
test_data, test_labels = company.test.next_batch()
test_labels = np.reshape(test_labels.T[1].T[:15],
(15, prediction_length))
test_data = np.reshape(test_data.T[1].T[:15],
(15, sequence_length))
predictions = sess.run(output,
feed_dict={
seq_input: test_data,
keep_prob: 1.0
})
predictions = np.reshape(predictions, (15, prediction_length))
labels, pred = [], []
for data, label, prediction in zip(test_data, test_labels,
predictions):
labels.append(np.concatenate((data, label), 0).tolist())
pred.append(np.concatenate((data, prediction), 0).tolist())
final_data, final_label = [], []
temp = 1
for d, l in zip(labels, pred):
final_data.append(d[0])
final_label.append(l[0])
final_data[:-1] += pred[temp - 1:][0]
final_label[:-1] += labels[temp - 1:][0]
temp += 1
# print temp
error = 0.05
resultfile = open("../resultfile.txt", "a")
if pred[0][len(pred[0]) - 2] > (pred[0][len(pred[0]) - 1] + error):
res_str = "CLOSING PRICE WILL GO UP FOR " + company_str + " >> BUY MORE SHARE..."
else:
res_str = "CLOSING PRICE WILL GO Down FOR " + company_str + " >> SELL MORE SHARE..."
resultfile.write(res_str)
resultfile.write("\n")
plt2.plot(test_labels.T[0][:-1], color='red', label='prediction')
plt2.plot(predictions.T[0][1:], color='blue', label='actual')
plt2.title(company_str)
plt2.xlabel('days')
plt2.ylabel('normalized closing prices')
plt2.legend(loc='upper left')
plt2.savefig("../graph/prediction" + company_str + ".png")
plt2.close()
datafile = open("../graphdata/" + company_str + "_data.js", "w")
labelfile = open("../graphdata/" + company_str + "_label.js", "w")
datafile.write("var " + company_str + "_CLOSING_PRICE_DATA = [ 0")
labelfile.write("var " + company_str + "_CLOSING_PRICE_LABEL = [ 0")
for dat, lab in zip(final_data[:-1], final_label[:-1]):
datafile.write(", '")
labelfile.write(", '")
datafile.write(str(dat))
labelfile.write(str(lab))
datafile.write("'")
labelfile.write("'")
datafile.write(" ];")
labelfile.write(" ];")
datafile.close()
labelfile.close()
print "Completed....."
names.append(name)
# found names to console
if len(names) > 0:
dayFlights = sort(flights[flights['localFlightDateStr']==localFlightDateStr],order=['points'])[::-1]
sortedNames = dayFlights[:]['name']
lastnames = ''
for name in sortedNames:
lastnames += '{} '.format(name.split(' ')[1])
print (lastnames)
else:
print('')
if not inPresent and mod(iSinceSave+1,nSaveInterval) == 0:
# '''save flights ; if unpublished, publish them.'''
publish(iflight, flights, notes, password, senderEmail, receiverEmail, adminEmail, inPresent, emailBlock)
save(flightsFile,iflight,flights)
writeFile('lastDateSearched.txt', loopDateStr)
iSinceSave = 0
print('Saved')
'''If loopDate is localToday, wait until tomorrow'''
while date.today() == loopDate: # wait until next morning. This ends for Australia at its midnight. It ends for USA at UTC's midnight.
inPresent = True
loopDate -= timedelta(days=1) # Come back to this date on sleep exit...incremented just after sleep.
publish(iflight, flights, notes, password, senderEmail, receiverEmail, adminEmail, inPresent, emailBlock,\
rankFlights, localAirports, groupName, googleGroupName, localeName, htmlFilePath)
save(flightsFile,iflight,flights)
yesterday = date.isoformat(date.today() - timedelta(days=1))
writeFile('lastDateSearched.txt', yesterday)
shutil.copy(flightsFile, '/media/sf_landscapes-zip/')
iSinceSave = 0
print('Saved')
"""Candle"""
candle_names = glob.glob("candle_in\*.png")
candle_names = sorted(candle_names)
candle_imgs = []
for img in candle_names:
candle_imgs.append(cv2.imread(img))
out1, out2, start, end = start_end_frame(candle_imgs)
candle_out = candle_imgs[start:end]
save_imgs(candle_out, "candle_out\\")
funcs.save(out1, "candle_out\out1.png")
funcs.save(out2, "candle_out\out2.png")
"""Personal"""
printer_names = glob.glob("printer_in\*.png")
printer_names = sorted(printer_names)
printer_imgs = []
for img in printer_names:
printer_imgs.append(cv2.imread(img))
out3, out4, startp, endp = start_end_frame(printer_imgs)
printer_out = printer_imgs[startp:endp]
save_imgs(printer_out, "printer_out\\")
funcs.save(out3, "printer_out\out3.png")