def setdefault(self, det):
self.p = np.array([])
self.cov = np.eye(4) * 1
self.num_p = 30
self.prob = np.array([])
self.det = det
# Init particles
self.init_p()
def setdefault(self, subject):
self.p = np.array([])
#self.cov = np.eye(5) * 1
self.cov = np.eye(6) * 1
self.num_p = 30
self.prob = np.array([])
self.det = subject
self.circle = subject.circle
self.rb = subject.rot_box
# Init particles
self.init_p()
def setdefault(self, subject):
self.p = np.array([])
#self.cov = np.eye(5) * 1
self.cov = np.eye(6) * 1
self.num_p = 30
self.prob = np.array([])
self.det = subject
self.circle = subject.circle
self.rb = subject.rot_box
# Init particles
self.init_p()
def alternativeAssociation(loss, distance, threshold, tr, res):
print '-----'
print 'Alfre2 association'
print '-----'
debug_flag = 0
if debug_flag: print loss
# get best detection for each tracker
y, x = loss.shape
new_res = np.array([[]])
for jj in range(y):
best_detection = loss[jj, :].argmin(axis=0)
s = loss[jj, best_detection]
# -- TOMA DE DECISIONES --
# 1) anomalia del score
mx = tr[jj].score_mu + 3 * tr[jj].score_sigma
mn = tr[jj].score_mu - 3 * tr[jj].score_sigma
if tr[jj].mature:
if s > mn and s < mx:
if not new_res.any():
new_res = np.array([[jj, best_detection]])
else:
new_res = np.vstack(
(new_res, np.array([jj, best_detection])))
else:
if not new_res.any():
new_res = np.array([[jj, best_detection]])
else:
new_res = np.vstack((new_res, np.array([jj, best_detection])))
if debug_flag:
#print 'loss: %s' \
# % loss[jj, best_detection]
print '[%s , %s] <-- %s' \
% (mn, mx, loss)
#print 'mu: %s, sigma: %s' \
# % (tr[jj].score_mu, tr[jj].score_sigma)
print tr[jj].score
print
print 'res'
print res
print 'new_res'
print new_res
return new_res
def alternativeAssociation(loss, distance, threshold, tr, res):
print '-----'
print 'Alfre2 association'
print '-----'
debug_flag=0
if debug_flag: print loss
# get best detection for each tracker
y, x = loss.shape
new_res = np.array([[]])
for jj in range(y):
best_detection = loss[jj, :].argmin(axis = 0)
s = loss[jj, best_detection]
# -- TOMA DE DECISIONES --
# 1) anomalia del score
mx = tr[jj].score_mu + 3*tr[jj].score_sigma
mn = tr[jj].score_mu - 3*tr[jj].score_sigma
if tr[jj].mature:
if s > mn and s < mx:
if not new_res.any():
new_res = np.array([[jj, best_detection]])
else:
new_res = np.vstack((new_res,np.array([jj, best_detection])))
else:
if not new_res.any():
new_res = np.array([[jj, best_detection]])
else:
new_res = np.vstack((new_res, np.array([jj, best_detection])))
if debug_flag:
#print 'loss: %s' \
# % loss[jj, best_detection]
print '[%s , %s] <-- %s' \
% (mn, mx, loss)
#print 'mu: %s, sigma: %s' \
# % (tr[jj].score_mu, tr[jj].score_sigma)
print tr[jj].score
print
print 'res'
print res
print 'new_res'
print new_res
return new_res
def getnotassociatedindex(len_sub, len_tr, del_tr, del_sub):
non_tr = []
non_sub = []
non_sub = np.array(range(len_sub))
non_tr = np.array(range(len_tr))
non_sub = np.delete(non_sub, del_sub)
non_tr = np.delete(non_tr, del_tr)
non_sub = non_sub.tolist()
non_tr = non_tr.tolist()
return non_sub, non_tr
def getnotassociatedindex(len_sub, len_tr, del_tr, del_sub):
non_tr = []
non_sub = []
non_sub = np.array(range(len_sub))
non_tr = np.array(range(len_tr))
non_sub = np.delete(non_sub, del_sub)
non_tr = np.delete(non_tr, del_tr)
non_sub = non_sub.tolist()
non_tr = non_tr.tolist()
return non_sub, non_tr
def plikelihood(self):
(x, y), (h, w), a = self.det
det = np.floor(np.array([[x, y, h, w]]))
yrep = np.ones((self.num_p, 4)) * det
R2 = np.sum(np.power(self.p - yrep, 2), 1)
width = 2 * (np.amax(np.sqrt(R2)) - np.amin(np.sqrt(R2)))
prob = np.exp(-R2 / width)
prob = prob / np.sum(prob)
self.prob = prob
self.sortprob()
def plikelihood(self):
if self.update:
(x, y), (h, w), a = self.rb
det = np.floor(np.array([[x, y, h, w, x - self.p_star[0], y - self.p_star[1]]]))
yrep = np.ones((self.num_p, 6)) * det
R2 = np.sum(np.power(self.p[:, 0:6] - yrep, 2), 1)
width = 2 * (np.amax(np.sqrt(R2)) - np.amin(np.sqrt(R2)))
prob = np.exp(- R2 / width)
a = np.sum(prob)
if a != 0.0:
prob = prob / np.sum(prob)
self.prob = prob
self.sortprob()
self.calculatepstar()
def plikelihood(self):
if self.update:
(x, y), (h, w), a = self.rb
det = np.floor(
np.array([[x, y, h, w, x - self.p_star[0],
y - self.p_star[1]]]))
yrep = np.ones((self.num_p, 6)) * det
R2 = np.sum(np.power(self.p[:, 0:6] - yrep, 2), 1)
width = 2 * (np.amax(np.sqrt(R2)) - np.amin(np.sqrt(R2)))
prob = np.exp(-R2 / width)
a = np.sum(prob)
if a != 0.0:
prob = prob / np.sum(prob)
self.prob = prob
self.sortprob()
self.calculatepstar()
def plikelihood_new(self):
if self.update:
(x, y), (h, w), a = self.rb
detectionExtended = np.floor(np.array([[x, y, h, w, x - self.p_star[0], y - self.p_star[1]]]))
yrep = np.ones((self.num_p, 6)) * detectionExtended
prob = gaussMixModelPDF(oneParticle,detectionExtended,p_star)
R2 = np.sum(np.power(self.p[:, 0:6] - yrep, 2), 1)
width = 2 * (np.amax(np.sqrt(R2)) - np.amin(np.sqrt(R2)))
prob = np.exp(- R2 / width)
a = np.sum(prob)
if a != 0.0:
prob = prob / np.sum(prob)
self.prob = prob
self.sortprob()
self.calculatepstar()
def plikelihood_new(self):
if self.update:
(x, y), (h, w), a = self.rb
detectionExtended = np.floor(
np.array([[x, y, h, w, x - self.p_star[0],
y - self.p_star[1]]]))
yrep = np.ones((self.num_p, 6)) * detectionExtended
prob = gaussMixModelPDF(oneParticle, detectionExtended, p_star)
R2 = np.sum(np.power(self.p[:, 0:6] - yrep, 2), 1)
width = 2 * (np.amax(np.sqrt(R2)) - np.amin(np.sqrt(R2)))
prob = np.exp(-R2 / width)
a = np.sum(prob)
if a != 0.0:
prob = prob / np.sum(prob)
self.prob = prob
self.sortprob()
self.calculatepstar()
