def main():
# initialization
pic = CorrectImage()
pic.add_path('data')
pic.add_image('initial.png')
pic.hough_transform(vary=False, plot=False) # set vary True to change edge filters, plot True for visualizations
pair = pic.line_pair(16)
slopes = map(pic.slope, pair)
loss_func = map(lambda s: loss(np.array(s[0]), np.array(s[1])), slopes)
# print pair
print loss_func
print health(loss_func, metric='arc_length'), '\n'
pop = Population()
pop.add_parents(pair, loss_func, 4)
pop.add_children(pic.mutation, p_mutate=0.01)
def main():
pop = Population()
pic = CorrectImage()
pic.add_path('data')
pic.add_image('initial.png')
pic.hough_transform(vary=False, plot=False) # set vary True to change edge filters, plot True for visualizations
pair = pic.line_pair(800)
total = 0
mean = []
avg_angle, count = 0, 0
for _ in xrange(500):
loss_func = evolve(pic, pop, pairs=pair, mutation_probability=0.01)
pop_health = health(loss_func, metric='arc_length')
total += pop_health
mean.append(total / (_ + 1))
pair = pop.population
if _ % 100 == 0:
print "Generation {0} health: \t{1}".format(_, pop_health), total / (_ + 1)
# draw(pair)
if (_ + 1) > 300:
count += 1
avg_angle += abs(np.mean(loss_func))
avg_angle /= count
print(avg_angle)
plt.clf()
plt.plot(range(500), mean)
plt.savefig('plots/generational_performance.png')
plt.show()
