def prepare_no_blood(name_scene='F(1)', id_exbl=24, index=0):
"""
performs the experiment from discussion:
number of blood pixels
data for Fig.10
parameters:
name_scene: name of the output image
id_exbl: id of the reference spectrum
index: iteration of the experiment
"""
data_scene, anno_scene = load_ds(name_scene)
blood_exbl = load_exbl(id_exbl)
X_bg = data_scene[anno_scene != 1]
X_blood_raw = data_scene[anno_scene == 1]
N_blood = len(X_blood_raw)
ratio_V = RATIOS_V
ratio_N = RATIOS_N
res = np.zeros((len(ratio_V), len(ratio_N)))
indices = np.arange(len(X_blood_raw))
np.random.shuffle(indices)
iii = 0
for i_v, ratio_vectors in enumerate(ratio_V):
for i_n, t_N in enumerate(ratio_N):
iii += 1
print(iii, ratio_vectors, t_N)
no_vectors = int(ratio_vectors * N_blood)
N = int(t_N * no_vectors) if t_N <= 1 else int(t_N)
X_blood = X_blood_raw[indices]
X_blood = X_blood[:no_vectors]
X = np.vstack((X_bg, X_blood))
y = np.concatenate(
(np.zeros(len(X_bg),
dtype=np.int32), np.ones(len(X_blood),
dtype=np.int32)))
blood = blood_exbl
mf = TwoStageMatchedFilter()
mf.fit(blood, X, N=N, N_supression=0)
pred_2 = mf.predict(X, stage='second')
vauc2, _, _ = comp_pr(pred_2, y)
res[i_v, i_n] = vauc2
np.savez_compressed('res/exp_no_blood_{}_{}'.format(name_scene, index),
res=res)
def show_table_row(name_frame, name_scene, code, N=1000, N_supression=0):
"""
plots rows of the result table (in LaTeX), Tab.3
parameters:
name_frame: name of the source image (inductive scenario)
name_scene: name of the output image
code: code for results file name
N: no. vectors in the 2nd stage as int or percentage string e.g. '33p'
N_supression: no. supressed vectors (extension of the algorithm, unused)
"""
str = ""
str += "{}&".format(code)
for i_f, f in enumerate(['scene', 'frame', 'exbl']):
fname = "res/{}_{}_{}_{}_{}.npz".format(name_frame, name_scene, N,
N_supression, f)
res = np.load(fname)
vauc, _, _ = comp_pr(res['pred_1'], res['anno'])
str += "{:0.2f}&".format(vauc)
if f == "exbl":
vauc, _, _ = comp_pr(res['pred_2'], res['anno'])
str += "{:0.2f}\\NN".format(vauc)
print(str)
def show_sensitivity_analysis(N_supression=0, is_pr=True, show=True):
"""
Graph of the impact of parameter N (no. vectors) , Fig.11
parameters:
N_supression: no. supressed vectors (extension of the algorithm, unused)
is_pr: use PR (True) or ROC (FALSE) AUC
show: True for plot.plot(), False for imsave
"""
images_n = [IMAGES[i] for i in [11, 4, 7, 8]]
markers = ['s', 'v', '<', '>', '1', '.', 'p']
plt.rcParams.update({'font.size': 14})
val_N = [
100, 250, 500, 750, 1000, 1250, 1500, 1750, 2000, 3000, 4000, 5000
]
for i_im, im in enumerate(images_n):
nn = []
for N in val_N:
name_frame = im['name_frame']
name_scene = im['name_scene']
fname = "res/{}_{}_{}_{}_{}.npz".format(name_frame, name_scene, N,
N_supression, "exbl")
res = np.load(fname)
pred = res['pred_2']
if is_pr:
vauc, _, _ = comp_pr(pred, res['anno'])
nn.append(vauc)
else:
vauc, _, _ = comp_roc(pred, res['anno'])
nn.append(vauc)
plt.plot(val_N,
nn,
label=im['code'],
marker=markers[i_im],
markevery=2)
if is_pr:
plt.ylabel('AUC(PR)')
else:
plt.ylabel('AUC(ROC)')
plt.xlabel('N')
plt.legend()
plt.tight_layout()
if show:
plt.show()
else:
plt.savefig('res/san_{}_{}.pdf'.format(N, "pr" if is_pr else "roc"),
bbox_inches='tight',
pad_inches=0)
plt.close()
def compare_in_columns(Ns=[750, 1000, '10p', '20p']):
"""
plots rows of the result table (in LaTeX), Tab.4
compare results for different values of N
parameters:
Ns: list values of N to plot for
"""
for im in IMAGES:
ss = "{}".format(im['code'].replace("_", "\_"))
for N in Ns:
fname = "res/{}_{}_{}_{}_{}.npz".format(im['name_frame'],
im['name_scene'], N, 0,
'exbl')
res = np.load(fname)
vauc, _, _ = comp_pr(res['pred_2'], res['anno'])
ss += "&{:0.2f}".format(vauc)
ss += '\\NN'
print(ss)
def show_stats(name_frame='F(1)',
name_scene='E(1)',
code='E(1)',
N=1000,
N_supression=0,
is_pr=True,
show=True):
"""
plots pr and roc curves for a given detection Fig.6-8
parameters:
name_frame: name of the source image (inductive scenario)
name_scene: name of the output image
code: code for results file name
N: no. vectors in the 2nd stage as int or percentage string e.g. '33p'
N_supression: no. supressed vectors (extension of the algorithm, unused)
is_pr: use PR (True) or ROC (FALSE) AUC
show: True for plot.plot(), False for imsave
"""
plt.rcParams.update({'font.size': 14})
markers = ['s', 'v', '<', '>', '1', '.', 'p']
labels = ['Ideal MF', 'Inductive MF', 'Algorithm 1']
for i_f, f in enumerate(['scene', 'frame', 'exbl']):
fname = "res/{}_{}_{}_{}_{}.npz".format(name_frame, name_scene, N,
N_supression, f)
print(fname)
res = np.load(fname)
pred = res['pred_2'] if f in ['exbl'] else res['pred_1']
if is_pr:
vauc, precision, recall = comp_pr(pred, res['anno'])
plt.plot(recall,
precision,
label="{} AUC:{:0.2f}".format(labels[i_f], vauc),
alpha=0.7,
marker=markers[i_f],
markevery=0.3,
lw=2)
else:
vauc, fpr, tpr = comp_roc(pred, res['anno'])
plt.plot(fpr,
tpr,
label="{} AUC:{:0.2f}".format(labels[i_f], vauc),
alpha=0.7,
marker=markers[i_f],
markevery=0.3,
lw=2)
if is_pr:
plt.ylabel('Precision')
plt.xlabel('Recall')
else:
plt.ylabel('TPR')
plt.xlabel('FPR')
plt.plot([0, 1], [0, 1], color='navy', lw=1, linestyle='--', alpha=0.1)
plt.legend()
plt.tight_layout()
if show:
plt.show()
else:
plt.savefig('res/res_{}_{}_{}.pdf'.format(code, N,
'pr' if is_pr else 'roc'),
bbox_inches='tight',
pad_inches=0)
plt.close()