Time the functions

This notebooks measures the runtime of each functionality.



In [1]:

    
import numpy as np
import cv2
import sys
import os
sys.path.insert(0, os.path.abspath('..'))

import salientregions as sr
import cProfile



In [2]:

    
%pylab inline









    



Populating the interactive namespace from numpy and matplotlib



In [3]:

    
#Load the image
path_to_image = 'images/graffiti.jpg'
img = cv2.imread(path_to_image)
sr.show_image(img)



In [4]:

    
%%timeit
#Time: creation of the detector
det = sr.SalientDetector(SE_size_factor=0.20,
                                lam_factor=4)









    



The slowest run took 29.09 times longer than the fastest. This could mean that an intermediate result is being cached.
100000 loops, best of 3: 1.61 µs per loop



In [5]:

    
det = sr.SalientDetector(SE_size_factor=0.20,
                                lam_factor=4)



In [6]:

    
%%timeit
#Time: detect all regions in color image
regions = det.detect(img,
                    find_holes=True,
                    find_islands=True,
                    find_indentations=True,
                    find_protrusions=True, 
                    visualize=False)









    



1 loop, best of 3: 5.15 s per loop



In [7]:

    
cProfile.run('det.detect(img, find_holes=True, find_islands=True, find_indentations=True, \
                    find_protrusions=True, visualize=False)')









    



         2132 function calls in 5.700 seconds

   Ordered by: standard name

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    0.001    0.001    5.700    5.700 <string>:1(<module>)
        3    0.000    0.000    0.000    0.000 _methods.py:25(_amax)
      480    0.000    0.000    0.003    0.000 _methods.py:31(_sum)
        1    0.007    0.007    1.237    1.237 binarization.py:143(binarize_withthreshold)
        1    0.000    0.000    1.237    1.237 binarization.py:229(binarize)
        1    0.000    0.000    0.000    0.000 binarydetector.py:112(reset)
        2    0.000    0.000    0.131    0.065 binarydetector.py:124(get_holes)
        2    0.000    0.000    0.323    0.161 binarydetector.py:136(get_islands)
        1    0.000    0.000    2.464    2.464 binarydetector.py:148(get_protrusions)
        1    0.000    0.000    1.540    1.540 binarydetector.py:157(get_indentations)
        2    0.000    0.000    0.449    0.225 binarydetector.py:166(_detect_holelike)
        2    0.082    0.041    4.005    2.002 binarydetector.py:189(_detect_protrusionlike)
        6    0.462    0.077    0.490    0.082 binarydetector.py:245(_remove_small_elements)
       42    0.010    0.000    0.063    0.002 binarydetector.py:300(_fill_image)
        1    0.000    0.000    0.000    0.000 binarydetector.py:46(__init__)
        1    0.000    0.000    4.458    4.458 binarydetector.py:60(detect)
        1    0.000    0.000    5.699    5.699 detectors.py:114(detect)
        1    0.000    0.000    0.001    0.001 detectors.py:45(detect)
        1    0.000    0.000    0.001    0.001 detectors.py:60(get_SE)
        2    0.000    0.000    0.000    0.000 {add}
        2    0.000    0.000    0.000    0.000 {bitwise_and}
       45    0.003    0.000    0.003    0.000 {bitwise_not}
        1    0.000    0.000    5.700    5.700 {built-in method builtins.exec}
        1    0.000    0.000    0.000    0.000 {built-in method builtins.len}
        1    0.000    0.000    0.000    0.000 {built-in method builtins.max}
        1    0.000    0.000    0.000    0.000 {built-in method builtins.min}
       43    0.033    0.001    0.033    0.001 {built-in method numpy.core.multiarray.array}
       91    0.005    0.000    0.005    0.000 {built-in method numpy.core.multiarray.zeros}
      170    1.256    0.007    1.256    0.007 {connectedComponentsWithStats}
        1    0.004    0.004    0.004    0.004 {cvtColor}
       42    0.043    0.001    0.043    0.001 {floodFill}
        1    0.001    0.001    0.001    0.001 {getStructuringElement}
        1    0.000    0.000    0.000    0.000 {method 'argmax' of 'numpy.ndarray' objects}
       48    0.003    0.000    0.003    0.000 {method 'copy' of 'numpy.ndarray' objects}
        1    0.000    0.000    0.000    0.000 {method 'disable' of '_lsprof.Profiler' objects}
        3    0.000    0.000    0.000    0.000 {method 'max' of 'numpy.ndarray' objects}
      483    0.003    0.000    0.003    0.000 {method 'reduce' of 'numpy.ufunc' objects}
      480    0.001    0.000    0.004    0.000 {method 'sum' of 'numpy.ndarray' objects}
        3    3.761    1.254    3.761    1.254 {morphologyEx}
      162    0.025    0.000    0.025    0.000 {threshold}



In [8]:

    
%%timeit
#Only holes and islands
regions = det.detect(img,
                    find_holes=True,
                    find_islands=True,
                    find_indentations=False,
                    find_protrusions=False,
                    visualize=False)









    



1 loop, best of 3: 1.32 s per loop

Binarization



In [9]:

    
lam_factor = 3
area_factor_large = 0.001
area_factor_verylarge = 0.1
lam = 50
connectivity = 4
weights=(0.33,0.33,0.33)
grayscale = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)



In [10]:

    
%%timeit
#Creation of the binarizer
binarizer = sr.DatadrivenBinarizer(area_factor_large=area_factor_large, area_factor_verylarge=area_factor_verylarge, 
                                           lam=lam, weights=weights, connectivity=connectivity)









    



The slowest run took 7.17 times longer than the fastest. This could mean that an intermediate result is being cached.
100000 loops, best of 3: 1.14 µs per loop



In [11]:

    
binarizer = sr.DatadrivenBinarizer(area_factor_large=area_factor_large, area_factor_verylarge=area_factor_verylarge, 
                                           lam=lam, weights=weights, connectivity=connectivity)



In [12]:

    
%%timeit
#The binarization
binarized = binarizer.binarize(grayscale, visualize=False)









    



1 loop, best of 3: 880 ms per loop



In [13]:

    
cProfile.run('binarizer.binarize(grayscale, visualize=False)')









    



         1782 function calls in 1.176 seconds

   Ordered by: standard name

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    0.000    0.000    1.176    1.176 <string>:1(<module>)
        3    0.000    0.000    0.000    0.000 _methods.py:25(_amax)
      480    0.000    0.000    0.003    0.000 _methods.py:31(_sum)
        1    0.010    0.010    1.176    1.176 binarization.py:143(binarize_withthreshold)
        1    0.000    0.000    1.176    1.176 binarization.py:229(binarize)
        1    0.000    0.000    1.176    1.176 {built-in method builtins.exec}
        1    0.000    0.000    0.000    0.000 {built-in method builtins.max}
        1    0.000    0.000    0.000    0.000 {built-in method builtins.min}
        3    0.002    0.001    0.002    0.001 {built-in method numpy.core.multiarray.zeros}
      160    1.136    0.007    1.136    0.007 {connectedComponentsWithStats}
        1    0.000    0.000    0.000    0.000 {method 'argmax' of 'numpy.ndarray' objects}
        1    0.000    0.000    0.000    0.000 {method 'disable' of '_lsprof.Profiler' objects}
        3    0.000    0.000    0.000    0.000 {method 'max' of 'numpy.ndarray' objects}
      483    0.003    0.000    0.003    0.000 {method 'reduce' of 'numpy.ufunc' objects}
      480    0.001    0.000    0.004    0.000 {method 'sum' of 'numpy.ndarray' objects}
      162    0.025    0.000    0.025    0.000 {threshold}

Binary detection



In [14]:

    
binarized = binarizer.binarize(grayscale, visualize=False)
regions = det.detect(img,
                    find_holes=True,
                    find_islands=True,
                    find_indentations=True,
                    find_protrusions=True, 
                    visualize=False)
se = det.SE



In [15]:

    
area_factor=0.05



In [16]:

    
%%timeit
detector = sr.BinaryDetector(se, lam, area_factor, connectivity)
regions = detector.detect(binarized, visualize=False)









    



1 loop, best of 3: 4.49 s per loop



In [17]:

    
detector = sr.BinaryDetector(se, lam, area_factor, connectivity)
cProfile.run('detector.detect(binarized, visualize=False)')









    



         1088 function calls in 5.229 seconds

   Ordered by: standard name

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    0.000    0.000    5.228    5.228 <string>:1(<module>)
        1    0.000    0.000    0.000    0.000 binarydetector.py:112(reset)
        2    0.000    0.000    0.261    0.131 binarydetector.py:124(get_holes)
        2    0.000    0.000    0.295    0.147 binarydetector.py:136(get_islands)
        1    0.000    0.000    2.939    2.939 binarydetector.py:148(get_protrusions)
        1    0.000    0.000    1.733    1.733 binarydetector.py:157(get_indentations)
        2    0.000    0.000    0.532    0.266 binarydetector.py:166(_detect_holelike)
        2    0.252    0.126    4.672    2.336 binarydetector.py:189(_detect_protrusionlike)
        6    0.532    0.089    0.571    0.095 binarydetector.py:245(_remove_small_elements)
      148    0.049    0.000    0.248    0.002 binarydetector.py:300(_fill_image)
        1    0.000    0.000    5.228    5.228 binarydetector.py:60(detect)
        2    0.000    0.000    0.000    0.000 {add}
        2    0.000    0.000    0.000    0.000 {bitwise_and}
      151    0.012    0.000    0.012    0.000 {bitwise_not}
        1    0.000    0.000    5.229    5.229 {built-in method builtins.exec}
      149    0.087    0.001    0.087    0.001 {built-in method numpy.core.multiarray.array}
      300    0.018    0.000    0.018    0.000 {built-in method numpy.core.multiarray.zeros}
       10    0.060    0.006    0.060    0.006 {connectedComponentsWithStats}
      148    0.160    0.001    0.160    0.001 {floodFill}
      154    0.010    0.000    0.010    0.000 {method 'copy' of 'numpy.ndarray' objects}
        1    0.000    0.000    0.000    0.000 {method 'disable' of '_lsprof.Profiler' objects}
        3    4.048    1.349    4.048    1.349 {morphologyEx}



In [18]:

    
#Only holes and islands
detector = sr.BinaryDetector(se, lam, area_factor, connectivity)
cProfile.run('detector.detect(binarized, find_indentations=False, \
                    find_protrusions=False, visualize=False)')









    



         32 function calls in 0.677 seconds

   Ordered by: standard name

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    0.000    0.000    0.676    0.676 <string>:1(<module>)
        1    0.000    0.000    0.000    0.000 binarydetector.py:112(reset)
        1    0.000    0.000    0.266    0.266 binarydetector.py:124(get_holes)
        1    0.000    0.000    0.410    0.410 binarydetector.py:136(get_islands)
        2    0.000    0.000    0.657    0.329 binarydetector.py:166(_detect_holelike)
        2    0.636    0.318    0.657    0.329 binarydetector.py:245(_remove_small_elements)
        2    0.001    0.001    0.019    0.009 binarydetector.py:300(_fill_image)
        1    0.000    0.000    0.676    0.676 binarydetector.py:60(detect)
        2    0.000    0.000    0.000    0.000 {bitwise_and}
        5    0.001    0.000    0.001    0.000 {bitwise_not}
        1    0.000    0.000    0.677    0.677 {built-in method builtins.exec}
        4    0.001    0.000    0.001    0.000 {built-in method numpy.core.multiarray.zeros}
        2    0.021    0.011    0.021    0.011 {connectedComponentsWithStats}
        2    0.015    0.007    0.015    0.007 {floodFill}
        4    0.001    0.000    0.001    0.000 {method 'copy' of 'numpy.ndarray' objects}
        1    0.000    0.000    0.000    0.000 {method 'disable' of '_lsprof.Profiler' objects}

MSER detection



In [19]:

    
mser = cv2.MSER_create()



In [20]:

    
%%timeit
regions = mser.detectRegions(img, None)









    



1 loop, best of 3: 433 ms per loop



In [21]:

    
cProfile.run('mser.detectRegions(img, None)')









    



         4 function calls in 0.680 seconds

   Ordered by: standard name

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    0.000    0.000    0.680    0.680 <string>:1(<module>)
        1    0.000    0.000    0.680    0.680 {built-in method builtins.exec}
        1    0.680    0.680    0.680    0.680 {method 'detectRegions' of 'cv2.MSER' objects}
        1    0.000    0.000    0.000    0.000 {method 'disable' of '_lsprof.Profiler' objects}

Conclusion

The tophat operation (for protrusions and indentations) is the bottleneck, takes 4.9 (of in total 5.7) seconds. The binarization is also quite slow because the function connectedComponentsWithStats is called for every threshold level.

The MSER detection is somewhat faster for a color image (about 2-3 times as fast).



In [22]:

    
%timeit cv2.morphologyEx(binarized, cv2.MORPH_TOPHAT, se)









    



1 loop, best of 3: 1.16 s per loop



In [23]:

    
%timeit cv2.morphologyEx(binarized, cv2.MORPH_OPEN, se)









    



1 loop, best of 3: 1.17 s per loop



In [24]:

    
%timeit cv2.erode(binarized, se)
%timeit cv2.dilate(binarized, se)









    



1 loop, best of 3: 583 ms per loop
1 loop, best of 3: 597 ms per loop



In [ ]: