charles/monoloco
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

remove unused figures

Browse Source
This commit is contained in:
lorenzo 2019-05-13 16:06:19 +02:00
parent 9c29ddb9f9
commit da4a351e95
1 changed files with 1 additions and 113 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

114
src/visuals/results.py
View File

@ -26,19 +26,6 @@ def print_results(dic_stats, show=False, save=False):
yy_gender = target_error(xx, mm_gender)
yy_gps = np.linspace(5., 5., xx.shape[0])
# # Recall results
# plt.figure(1)
# plt.xlabel("Distance [meters]")
# plt.ylabel("Detected instances")
# plt.xlim(x_min, x_max)
# for method in ['our', 'm3d', '3dop']:
# xxs = get_distances(clusters)
# dic_cnt = dic_stats[phase][method]['cnt']
# cnts = get_values(dic_cnt, clusters)
# plt.plot(xxs, cnts, marker='s', label=method + '_method')
# plt.legend()
# plt.show()
#
# Precision on same instances
fig_name = 'results.png'
plt.xlabel("Distance [meters]")
@ -158,103 +145,4 @@ def get_confidence_points(confidences, distances, errors):
distance_points.append(dd)
distance_points.append(dd)
return distance_points, confidence_points
# def histogram(self):
# """
# Visualize histograms to compare error performances of net and baseline
# """
# # for mode in ['mean', 'std']:
# for mode in ['mean']:
#
# # for phase in ['train', 'val', 'test']:
# for phase in ['test']:
#
# err_geom = []
# err_net = []
# counts = []
# clusters = ('10', '20', '30', '40', '50', '>50', 'all')
# for clst in clusters:
#
# err_geom.append(dic_geom[phase]['error'][clst][mode])
# counts.append(dic_geom[phase]['error'][clst]['count'])
# err_net.append(dic_net[phase][clst][mode])
#
# nn = len(clusters)
#
# # TODO Shortcut to print 2 models in a single histogram
# err_net_l1 = [1.1, 1.32, 2.19, 3.29, 4.38, 6.58, 2.21]
#
# ind = np.arange(nn) # the x locations for the groups
# width = 0.30 # the width of the bars
#
# fig, ax = plt.subplots()
# rects1 = ax.bar(ind, err_geom, width, color='b')
# rects2 = ax.bar(ind + width, err_net_l1, width, color='r')
# rects3 = ax.bar(ind + 2 * width, err_net, width, color='g')
#
# # add some text for labels, title and axes ticks
# ax.set_ylabel('Distance error [m]')
# ax.set_title(mode + ' of errors in ' + phase)
# ax.set_xticks(ind + width / 2)
# ax.set_xticklabels(clusters)
#
# ax.legend((rects1[0], rects2[0], rects3[0]), ('Geometrical', 'L1 Loss', 'Laplacian Loss'))
#
# # Attach a text label above each bar displaying number of annotations
# for idx, rect in enumerate(rects1):
# height = rect.get_height()
# count = counts[idx]
# ax.text(rect.get_x() + rect.get_width() / 2., 1.05 * height,
# '%d' % int(count), ha='center', va='bottom')
#
# plt.show()
# def error(self):
# """
# Visualize error as function of the distance on the test set and compare it with target errors based on human
# height analyses. We consider:
# Position error in meters due to a height variation of 7 cm (Standard deviation already knowing the sex)
# Position error not knowing the gender (13cm as average difference --> 7.5cm of error to add)
# """
# phase = 'test'
# xx = np.linspace(0, 60, 100)
# mm_std = 0.04
# mm_gender = 0.0556
# clusters = tuple([clst for clst in dic_net[phase] if clst not in ['all', '>50']])
# # errors_geom = [dic_geom[phase]['error'][clst][mode] for clst in clusters]
# errors_net = [dic_net[phase][clst]['mean'] for clst in clusters]
# confidences = [dic_net[phase][clst]['std'][0] for clst in clusters]
# distances = get_distances(clusters)
# dds, zzs = get_confidence_points(confidences, distances, errors_net)
#
# # Set the plot
# fig = plt.figure()
# ax = plt.subplot(111)
# plt.xlim(0, 50)
# plt.xlabel("Distance [meters]")
# plt.ylabel("Error [meters]")
# plt.title("Error on Z Position Estimate due to height variation")
#
# # Plot the target errors
# yy_std = target_error(xx, mm_std)
# yy_gender = target_error(xx, mm_gender)
# # yy_child = target_error(xx, mm_child)
# # yy_test = target_error(xx, mm_test)
# yy_gps = np.linspace(5., 5., xx.shape[0])
# plt.plot(xx, yy_std, '--', label="Knowing the gender", color='g')
# plt.plot(xx, yy_gender, '--', label="NOT knowing the gender", color='b')
# plt.plot(xx, yy_gps, '-', label="GPS Error", color='y')
#
# for idx in range(0, len(dds), 2):
#
# plt.plot(dds[idx: idx+2], zzs[idx: idx+2], color='k')
#
# # Plot the geometric and net errors as dots
# _ = ax.plot(distances, errors_net, 'ro', marker='s', label="Network Error (test)", markersize='8')
# # _ = ax.plot(distances, errors_geom, 'ko', marker='o', label="Baseline Error", markersize='5')
#
# ax.legend(loc='upper left')
# plt.show()
return distance_points, confidence_points