def start_detect(self):
if self.executor is not None:
self.executor[1]()
self.executor = None
self.proc["text"] = "Detect"
return
cpus = int(self.cpu_sel.get())
raw_ar = self.ar.get().split(":")
if len(raw_ar) != 2:
showerror("Invalid data", "Invalid aspect ratio", master=self)
return
try:
ar = tuple(int(a) for a in raw_ar)
except ValueError:
showerror("Invalid data", "Invalid aspect ratio", master=self)
return
try:
from_ = self.from_.get()
except ValueError:
showerror("Invalid data", "Invalid start frame", master=self)
return
try:
to = self.to.get()
except ValueError:
showerror("Invalid data", "Invalid end frame", master=self)
return
self.proc["text"] = "Stop"
self.progress["max"] = len(xrange(from_, to + 1, 2)) * len(
Tester.testers)
self.progress["value"] = 0
self.reset_plots(from_, to)
self.reset_data()
executor = Executor(self.filename,
range(from_, to + 1, 2),
self.framecb(),
aspect_ratio=ar,
cpus=cpus)
cb, stop = executor.test()
self.after(1, self.sync)
self.executor = (executor, stop, cb)
def start_detect(self):
if self.executor is not None:
self.executor[1]()
self.executor = None
self.proc["text"] = "Detect"
return
cpus = int(self.cpu_sel.get())
raw_ar = self.ar.get().split(":")
if len(raw_ar) != 2:
showerror("Invalid data", "Invalid aspect ratio", master=self)
return
try:
ar = tuple(int(a) for a in raw_ar)
except ValueError:
showerror("Invalid data", "Invalid aspect ratio", master=self)
return
try:
from_ = self.from_.get()
except ValueError:
showerror("Invalid data", "Invalid start frame", master=self)
return
try:
to = self.to.get()
except ValueError:
showerror("Invalid data", "Invalid end frame", master=self)
return
self.proc["text"] = "Stop"
self.progress["max"] = len(xrange(from_, to+1, 2))*len(Tester.testers)
self.progress["value"] = 0
self.reset_plots(from_, to)
self.reset_data()
executor = Executor(self.filename, range(from_, to+1, 2), self.framecb(), aspect_ratio=ar, cpus=cpus)
cb, stop = executor.test()
self.after(1, self.sync)
self.executor = (executor, stop, cb)
def run():
vars = docopt.docopt(__doc__)
fstep = range(*(int(i) for i in vars["FRAMES"].split(":")))
hstep = range(*(int(i) for i in vars["HEIGHTS"].split(":")))
src = vars["SRC"]
app = Executor(src, hstep, fstep)
cb, stop = app.test()
class _data(object):
testers = {}
res_before = {}
lock = threading.Lock()
@classmethod
def init(cls):
for name, tester in Tester.testers.items():
data = [[], [], []]
cls.testers[name] = data
plots, names = [],[]
for name, tester in Tester.testers.items():
pyplot.subplot(211)
plt = pyplot.scatter([], [], color=tester.color)
plots.append(plt)
names.append(name)
pyplot.grid()
pyplot.legend(plots, names, loc='upper left', bbox_to_anchor=(0, 1))
plots, names = [],[]
for name, tester in Tester.testers.items():
pyplot.subplot(212)
plt = pyplot.scatter([], [], color=tester.color)
plots.append(plt)
names.append(name)
pyplot.grid()
pyplot.legend(plots, names, loc='upper left', bbox_to_anchor=(0, 1))
@classmethod
def on_new_item(cls, tester, width, height, frame, result, image):
if tester not in cls.res_before:
cls.res_before[tester] = result
return
diff = result - cls.res_before[tester]
heights, results, actual = cls.testers[tester]
with cls.lock:
heights.append(height)
results.append(diff)
actual.append(result)
cls.res_before[tester] = result
def datathread():
while cb(_data.on_new_item):
pass
_data.init()
thread = threading.Thread(target=datathread)
thread.daemon = True
thread.start()
def _update():
pyplot.subplot(211)
plots, names = [], []
for name, tester in Tester.testers.items():
with _data.lock:
heights, delta, results = _data.testers[name]
plt = pyplot.scatter(heights, delta, color=tester.color, marker="x")
plots.append(plt)
names.append(name)
pyplot.title("Resolution Rescaling Image Comparison [d/dx]")
pyplot.legend(plots, names, loc='upper left', bbox_to_anchor=(0, 1))
pyplot.subplot(212)
plots, names = [], []
for name, tester in Tester.testers.items():
with _data.lock:
heights, delta, results = _data.testers[name]
plt = pyplot.scatter(heights, results, color=tester.color)
plots.append(plt)
names.append(name)
pyplot.title("Resolution Rescaling Image Comparison")
pyplot.legend(plots, names, loc='upper left', bbox_to_anchor=(0, 1))
pyplot.draw()
pyplot.subplot(211)
pyplot.grid()
pyplot.subplot(212)
pyplot.grid()
tstart = time.time()
while thread.is_alive():
if time.time()-tstart > 5:
_update()
tstart = time.time()
pyplot.pause(.5)
_update()
stop()
import numpy as np
def is_outlier(points, thresh=5):
"""
Returns a boolean array with True if points are outliers and False
otherwise.
Parameters:
-----------
points : An numobservations by numdimensions array of observations
thresh : The modified z-score to use as a threshold. Observations with
a modified z-score (based on the median absolute deviation) greater
than this value will be classified as outliers.
Returns:
--------
mask : A numobservations-length boolean array.
References:
----------
Boris Iglewicz and David Hoaglin (1993), "Volume 16: How to Detect and
Handle Outliers", The ASQC Basic References in Quality Control:
Statistical Techniques, Edward F. Mykytka, Ph.D., Editor.
"""
if len(points.shape) == 1:
points = points[:,None]
median = np.median(points, axis=0)
diff = np.sum((points - median)**2, axis=-1)
diff = np.sqrt(diff)
med_abs_deviation = np.median(diff)
modified_z_score = 0.6745 * diff / med_abs_deviation
return modified_z_score > thresh
for name, tester in Tester.testers.items():
heights, results, _ = _data.testers[name]
results = np.array(results)
mask = is_outlier(results)
outliers = np.array([results, np.arange(len(results), dtype=np.uint16)]).T[mask]
ibefore = None
for val, i in outliers:
if ibefore is None:
ibefore = (i, val)
continue
if i - ibefore[0] > 1:
ibefore = (i, val)
continue
if ibefore[1] < 0:
ibefore = (i, val)
continue
print("Possible interesting point at height: %d@%s" % (heights[int(ibefore[0])], name))
ibefore = (i, val)
pyplot.show()