def results_evaluation(results, test_GT, prints=True):
# Evaluation sklearn
if prints:
sys.stdout.write('Evaluating results... ')
t = time.time()
metrics = ev.getMetrics(test_GT, results)
if prints:
elapsed = time.time() - t
sys.stdout.write(str(elapsed) + ' sec \n\n')
print "Recall: " + str(metrics[0] * 100)
print "Precision: " + str(metrics[1] * 100)
print "F1: " + str(metrics[2] * 100)
return metrics
def single_execution(train, test, test_GT, alpha, ro):
sys.stdout.write('Computing background substraction... ')
# Background substraction
g = GaussianModelling(alpha=alpha, adaptive_ratio=ro)
g.fit(train)
results = g.predict(test)
# Evaluation sklearn
t = time.time()
metrics = ev.getMetrics(test_GT, results)
elapsed = time.time() - t
sys.stdout.write(str(elapsed) + ' sec \n')
print "Recall: " + str(metrics[0] * 100)
print "Precision: " + str(metrics[1] * 100)
print "F1: " + str(metrics[2] * 100)
def grid_search(train, test, test_GT):
alpha_range = np.around(np.arange(2.5, 4.52, 0.1), decimals=2)
ro_range = np.around(np.arange(0, 0.4, 0.05), decimals=2)
f1_matrix = np.zeros([len(alpha_range), len(ro_range)])
for i in range(len(alpha_range)):
alpha = alpha_range[i]
for j in range(len(ro_range)):
ro = ro_range[j]
sys.stdout.write("(alpha=" + str(alpha) + ", ro=" + str(ro) + ") ")
# Background substraction
g = GaussianModelling(alpha=alpha, adaptive_ratio=ro)
g.fit(train)
results = g.predict(test)
# Evaluation sklearn
t = time.time()
metrics = ev.getMetrics(test_GT, results)
elapsed = time.time() - t
sys.stdout.write(str(elapsed) + ' sec \n')
f1_matrix[i, j] = metrics[2]
# Plot grid search
X, Y = np.meshgrid(ro_range, alpha_range)
Z = f1_matrix
f1_max = np.max(f1_matrix)
f1_max_idx = np.argmax(f1_matrix)
best_alpha = Y.flatten()[f1_max_idx]
best_ro = X.flatten()[f1_max_idx]
print "F1: " + str(np.around(f1_max, decimals=5)) + " (alpha=" + str(
best_alpha) + ", ro=" + str(best_ro) + ")"
fig = plt.figure()
ax = fig.gca(projection='3d')
ax.plot_surface(X, Y, Z, cmap='plasma')
axis = ["Ro", "Alpha", "F1-score"]
ax.set_xlabel(axis[0])
ax.set_ylabel(axis[1])
ax.set_zlabel(axis[2])
# plt.savefig('grid_search.png',dpi=300)
plt.show()
def f1score_alpha(train, test, test_GT):
# Task 1.2 - F1-score vs Alpha
alpha_range = np.around(np.arange(0, 6, 0.1), decimals=2)
metrics_array = []
for alpha in alpha_range:
sys.stdout.write("(alpha=" + str(np.around(alpha, decimals=2)) + ") ")
# Background substraction
g = GaussianModelling(alpha=alpha)
g.fit(train)
results = g.predict(test)
# Evaluation sklearn
t = time.time()
metrics = ev.getMetrics(test_GT, results)
elapsed = time.time() - t
sys.stdout.write(str(elapsed) + ' sec \n')
metrics_array.append(metrics)
metrics_array = np.array(metrics_array)
return metrics_array[:, 2]
def f1score_alpha(train, test, test_GT):
# Task 1.2 - F1-score vs Alpha
alpha_range = np.around(np.arange(1.5, 2.5, 0.1), decimals=2)
metrics_array = []
for alpha in alpha_range:
sys.stdout.write("(alpha=" + str(np.around(alpha, decimals=2)) + ") ")
# Background substraction
g = GaussianModelling(alpha=alpha)
g.fit(train)
results = g.predict(test)
# Evaluation sklearn
t = time.time()
metrics = ev.getMetrics(test_GT, results)
elapsed = time.time() - t
sys.stdout.write(str(elapsed) + ' sec \n')
metrics_array.append(metrics)
# TASK 2.2 - Plot F1-score vs Alpha
x = [alpha_range, alpha_range, alpha_range]
metrics_array = np.array(metrics_array)
y = [metrics_array[:, 0], metrics_array[:, 1], metrics_array[:, 2]]
f1_max = np.max(metrics_array[:, 2])
f1_max_idx = np.argmax(metrics_array[:, 2])
best_alpha = alpha_range[f1_max_idx]
print "F1: " + str(np.around(
f1_max, decimals=4)) + " (alpha=" + str(best_alpha) + ")"
axis = ["Alpha", "F1-score"]
labels = ["Precision", "Recall", "F1"]
ev.plotGraphics(x, y, axis, labels)
def main():
# Read dataset
#dataset = Dataset('highway',1050, 1350)
#dataset = Dataset('fall', 1461, 1560)
dataset = Dataset('traffic', 951, 1050)
imgs = dataset.readInput()
imgs_GT = dataset.readGT()
# cap = cv2.VideoCapture('source.mp4')
kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (3, 3))
fgbg = createBackgroundSubtractor()
substracted = []
gif_frames = []
fig = plt.figure()
for i in range(len(imgs)):
frame = imgs[i]
fgmask = fgbg.apply(frame)
fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_OPEN, kernel)
substracted.append(fgmask)
if gif is True:
g = plt.imshow(fgmask, cmap='gray', animated=True)
plt.axis("off")
gif_frames.append([g])
if showFrames is True:
cv2.imshow('input', imgs[i])
cv2.imshow('frame', fgmask)
k = cv2.waitKey(30) & 0xff
if k == 27:
break
if PR_curve is True:
precision, recall, auc_val = ev.getPR_AUC(imgs_GT, substracted)
plt.step(recall, precision, color='b', alpha=0.2, where='post')
plt.fill_between(recall, precision, step='post', alpha=0.2, color='g')
plt.xlabel('Recall')
plt.ylabel('Precision')
plt.ylim([0.0, 1.0])
plt.xlim([0.0, 1.0])
plt.title("Precision-Recall curve - Fall - MOG2")
plt.show()
print len(precision)
print 'precision'
print sum(precision) / len(precision)
print 'recall'
print recall
print 'Area under the curve '
print auc_val
if Metrics is True:
metrics = ev.getMetrics(imgs_GT, substracted)
print metrics
if gif is True:
anim = animation.ArtistAnimation(fig,
gif_frames,
interval=len(imgs),
blit=True)
anim.save('animation.gif', writer='imagemagick', fps=10)
# plt.show()
#cap.release()
cv2.destroyAllWindows()