def showPNGBad(grid):
import matplotlib.pyplot as plt
"""Generate a simple image of the maze."""
plt.figure(figsize=(10, 5))
plt.imshow(grid, cmap=plt.cm.binary, interpolation="nearest")
plt.xticks([]), plt.yticks([])
plt.save_fig(format="png")
def plot_error_types():
types_dict = json.load(open("/Users/shankari/cluster_ground_truth/smoothing/other_auto/smoothing_categories"))
types_len_dict = {}
for (category, cat_list) in types_dict.iteritems:
types_len_dict[category] = len(cat_list)
plt.xlabel("Number of sections")
plt.ylabel("Category")
plt.tight_layout()
plt.save_fig("/tmp/smoothing_category_plot.png")
def plot_error_types():
types_dict = json.load(
open(
"/Users/shankari/cluster_ground_truth/smoothing/other_auto/smoothing_categories"
))
types_len_dict = {}
for (category, cat_list) in types_dict.iteritems:
types_len_dict[category] = len(cat_list)
plt.xlabel("Number of sections")
plt.ylabel("Category")
plt.tight_layout()
plt.save_fig("/tmp/smoothing_category_plot.png")
def plot_history(self, history):
N = np.arange(0, self.NUM_EPOCHS)
plt.figure()
plt.plot(N, history.history['loss'], label='train_loss')
plt.plot(N, history.history['val_loss'], label='validation_loss')
plt.plot(N, history.history['acc'], label='train_acc')
plt.plot(N, history.history['val_acc'], label='validation_acc')
plt.title('Training History')
plt.xlabel('Epoch')
plt.ylabel('Loss / Accuracy')
plt.legend(loc='upper right')
plt.save_fig(self.model_path + '/training_result.png')
def process_network(G, namespace):
print 'Nodes:', len(G)
print 'Edges:', G.number_of_edges()
if namespace.clustering_coefficient:
print 'Clustering Coefficient:', nx.average_clustering(G)
if namespace.components:
components = nx.connected_component_subgraphs(G)
print 'Number of Components:', len(components)
isles = [c for c in components if len(c) == 1]
print 'Isles:', len(isles)
print 'Largest Component Size:', len(components[0])
else:
components = None
if namespace.cpl:
if namespace.approximate_cpl:
average_shortest_path_length = approximate_cpl
else:
print 'Using full slow CPL'
average_shortest_path_length = nx.average_shortest_path_length
if components is None:
components = nx.connected_component_subgraphs(G)
for i, g in enumerate(
g for g in components
if float(len(g)) / float(len(G)) > namespace.component_size):
print 'CPL %d: (%f)' % (i, float(len(g)) / float(len(G)))
print average_shortest_path_length(g)
if namespace.assortativity:
print 'Assortativity: NOT IMPLEMENTED.'
if namespace.degree_distribution:
hst = nx.degree_histogram(G)
plt.subplot(121)
plt.xscale('log')
plt.yscale('log')
plt.title("Degree Distribution")
plt.ylabel("Occurrencies")
plt.xlabel("Degree")
plt.plot(range(len(hst)), hst, marker='+')
plt.subplot(122)
ccdf = pynetsym.mathutil.ccdf(hst)
plt.xscale('log')
plt.yscale('log')
plt.title("CCDF Degree Distribution")
plt.ylabel("$P(X>x)$")
plt.xlabel("Degree")
plt.plot(range(len(ccdf)), ccdf, color='red')
if namespace.degree_distribution_out is None:
plt.show()
else:
plt.save_fig(namespace.degree_distribution_out)
def classify(x_train, y_train, x_test, classifier):
'''
from rachel's pipeline_library
'''
model = classifier
model.fit(x_train, y_train)
if str(classifier).split('(')[0] == 'LinearSVC':
predicted_scores = model.decision_function(x_test)
else:
predicted_scores = model.predict_proba(x_test)[:, 1]
plt.hist(predicted_scores)
plt.save_fig()
return predicted_scores
def make_plot():
with open('i2.dump', 'rb') as dump_file:
polies = pickle.load(dump_file)
plt = GeoPlot()
plt.add_collection(
PatchCollection(polies,
alpha=0.5,
edgecolor='black',
linewidth=0.001,
match_original=True))
plt.set_xlim(-120, -40)
plt.set_ylim(-15, 80)
plt.save_fig('../figures/i2.pdf')
def make_heatmap(sim,names=None,outname=None):
fig = plt.figure()
ax = sns.heatmap(sim)
if names is not None:
ax.set_xticklabels(names,rotation=90)
ax.set_yticklabels(names[::-1],rotation=0)
plt.title('Cosine Similarity of Sound Features')
plt.tight_layout()
if outname:
plt.save_fig(outname)
return
plt.show()
return
def process_network(G, namespace):
print 'Nodes:', len(G)
print 'Edges:', G.number_of_edges()
if namespace.clustering_coefficient:
print 'Clustering Coefficient:', nx.average_clustering(G)
if namespace.components:
components = nx.connected_component_subgraphs(G)
print 'Number of Components:', len(components)
isles = [c for c in components if len(c) == 1]
print 'Isles:', len(isles)
print 'Largest Component Size:', len(components[0])
else: components = None
if namespace.cpl:
if namespace.approximate_cpl:
average_shortest_path_length = approximate_cpl
else:
print 'Using full slow CPL'
average_shortest_path_length = nx.average_shortest_path_length
if components is None:
components = nx.connected_component_subgraphs(G)
for i, g in enumerate(g for g in components if
float(len(g))/float(len(G)) >
namespace.component_size):
print 'CPL %d: (%f)' % (i, float(len(g))/float(len(G)))
print average_shortest_path_length(g)
if namespace.assortativity:
print 'Assortativity: NOT IMPLEMENTED.'
if namespace.degree_distribution:
hst = nx.degree_histogram(G)
plt.subplot(121)
plt.xscale('log')
plt.yscale('log')
plt.title("Degree Distribution")
plt.ylabel("Occurrencies")
plt.xlabel("Degree")
plt.plot(range(len(hst)), hst, marker='+')
plt.subplot(122)
ccdf = pynetsym.mathutil.ccdf(hst)
plt.xscale('log')
plt.yscale('log')
plt.title("CCDF Degree Distribution")
plt.ylabel("$P(X>x)$")
plt.xlabel("Degree")
plt.plot(range(len(ccdf)), ccdf, color='red')
if namespace.degree_distribution_out is None:
plt.show()
else:
plt.save_fig(namespace.degree_distribution_out)
def __init__(self):
root = '.data'
train_data = torchvision.datasets.CIFAR10(root,
train=True,
transform=None,
target_transform=None,
download=True)
test_data = torchvision.datasets.CIFAR10(root,
train=False,
transform=None,
target_transform=None,
download=True)
print(len(train_data), len(test_data))
CLUSTERS_PATH = '.data/clusters.pkl'
if not os.path.exists(CLUSTERS_PATH):
# get random 5 pixels per image and stack them all up as rgb values to get half a million random pixels
pluck_rgb = lambda x: torch.from_numpy(np.array(x)).view(
32 * 32, 3)[torch.randperm(32 * 32)[:5], :]
px = torch.cat([pluck_rgb(x) for x, y in train_data],
dim=0).float()
print(px.size())
ncluster = 512
with torch.no_grad():
C = ImageDataset.kmeans(px, ncluster, niter=8)
with open(CLUSTERS_PATH, 'wb') as f:
pickle.dump(C, f)
else:
with open(CLUSTERS_PATH, 'rb') as f:
C = pickle.load(f)
# encode the training examples with our codebook to visualize how much we've lost in the discretization
n_samples = 16
ncol = 8
nrow = n_samples // ncol + 1
plt.figure(figsize=(20, 10))
for i in range(n_samples):
# encode and decode random data
x, y = train_data[np.random.randint(0, len(train_data))]
xpt = torch.from_numpy(np.array(x)).float().view(32 * 32, 3)
ix = ((xpt[:, None, :] - C[None, :, :])**2).sum(-1).argmin(
1) # cluster assignments for each pixel
# these images should look normal ideally
plt.subplot(nrow, ncol, i + 1)
plt.plot(C[ix].view(32, 32, 3).numpy().astype(np.uint8))
plt.save_fig('temp.png')
def plot_spectrum():
spect_file=np.load(spect_file_path + '.npy')
energy_centers=spect_file[::,0]
spect_medians={}
keys=spect_file[0,1].keys()
print 'The keys: ', keys
for key in keys:
spect_medians[key]=np.array([spect_file[::,1][en][key][1] for en in range(len(energy_centers))])
for key in keys:
plt.plot(energy_centers,spect_medians[key],label=key)
plt.xscale('log')
plt.yscale('log')
plt.ylim([1e-8,1e-4]);
plt.ylabel('$E^2 dN / dE$ [ GeV / cm$^2$ / s / sr]',fontsize=16)
plt.xlabel('E [ GeV ]',fontsize=16)
plt.legend()
plt.save_fig(b.plots_dir + spect_plot_tag + '.pdf')
def hists(files,legend,**kwargs): #plot histogram of RSSIs from each file
fig, ax=plt.subplots()
print('avgs:')
cmap=plt.cm.get_cmap('tab20')
i=0
k1={**kwargs}
k1.pop('adver',None)
for f in files:
r=rssis(f,**kwargs)
print(f+':'+str(np.mean(r)))
ax.hist(r,label=' '+f,color=cmap(i/20),bins=np.max(r)-np.min(r)+1,**k1)
i+=1
if legend:
ax.legend()
ax.set_xlabel('RSSIs')
plt.show()
plt.save_fig(DA_+'tmp.png')
def Multi_KNN_Class(X,
y,
test_s=0.1,
neighbors=[1, 2, 3],
p_val=[1, 2],
leaf=[30],
iterations=20,
fig_s=(15, 9),
path=os.getcwd(),
plot=False,
verbose=True):
"""test out all combinations of hyperparameters to find the best model configuration. Returns statistics for mean and standard
deviation of accuracy over the amount of iterations for each hyperparameter settings."""
mu_sigma_list = []
for s in list(set(itertools.product(neighbors, p_val, leaf))):
i, j, k = s
acc_list = []
knn = KNeighborsClassifier(n_neighbors=i, p=j, leaf_size=k)
for r in range(iterations):
X_train, X_val, y_train, y_val = train_test_split(X,
y,
test_size=test_s)
knn.fit(X_train, y_train)
acc = knn.score(X_val, y_val)
acc_list.append(acc)
mu = np.mean(acc_list)
sigma = np.std(acc_list)
mu_sigma_list.append(('{}_NN__p_{}__leafsize_{}'.format(i, j,
k), mu, sigma))
if verbose:
print('{}_NN__p_{}__leafsize_{}'.format(i, j, k), mu, sigma)
if plot:
x_axis = list(range(len(acc_list)))
plt.figure(figsize=fig_s)
text = 'Accuracy: {}_NN__p_{}__leafsize_{}'.format(i, j, k)
plt.title(text)
plt.plot(x_axis, acc_list)
plt.save_fig(path + '/{}.png'.format(text))
return mu_sigma_list
def run_training(model, data, optimizer, loss_function, epochs, generator):
t_loss = []
e_loss = []
for epoch in range(epochs):
train_loss = train(model, data(0, val_split), optimizer, loss_function, epoch)
eval_loss = evaluate(model, data(val_split, 1), loss_function, epoch)
t_plot = np.asarray(t_loss.append(train_loss))
e_plot = np.asarray(e_loss.append(eval_loss))
t_plot = np.reshape(-1, t_plot.shape[-1])
e_plot = np.reshape(-1, e_plot.shape[-1])
iters = np.arange(0, t_plot.shape[0]*100, 100)
plt.plot(t_plot, iters, 'r', e_plot, 'b', linewidth=2.0)
plt.save_fig(params['plot_save_path'])
plt.clf()
if epoch%10 == 0:
save_checkpoints(model, optim, epoch, params['save_dir'])
#TODO: Setup a random spectrogram and save_path
spec2wav(generator, model, spectrogram, save_path)
times.append(frame_number)
if (zeros/nonzeros) <= proportion_dark:
newfmf.add_frame(frame, timestamp)
sumframe = sumframe + frame
listynew.append(zeros/nonzeros)
timesnew.append(frame_number)
else:
pass
newfmf.close()
#plot proportion of black pixels before and after
plt.plot(times, listy, color='b')
plt.plot(timesnew, listynew, color='r')
plt.save_fig(sys.argv[2] + '_zeros.svg')
plt.show()
#plot average intensity (vignette)
vignette = sumframe / len(timesnew)
plt.imshow(vignette)
plt.save_fig(sys.argv[2] + '_vignette.svg')
plt.show()
#Save vignette values
DataFrame(vignette).to_csv(sys.argv[2] + '_array.csv', sep=',')
print 'done.'
filenames = glob.glob(config.savedir + config.dataname + '*.hd5')
print(filenames)
dataset = functions.DaskArray_hd5loadDataset(filenames, verbose=config.verbose)
print('dataset loaded with keys:')
print(dataset.keys())
Y = dataset['Y']
data = []
for pipeline in dataset.keys():
if pipeline in config.pipelines:
data.append(dataset[pipeline])
if config.visualize == True:
for j in range(5):
i = random.randint(0, dataset[pipeline].shape[0])
plt.title(pipeline + str(Y[i].compute()))
plt.plot(dataset[pipeline][i, :].compute())
plt.save_fig('./figures/' + pipeline +
str(Y[i].compute() + '.png'))
X = functions.da.concatenate(data, axis=1)
######################################make NN ################################################################
scaler = StandardScaler(copy=True)
le = LabelEncoder()
enc = OneHotEncoder(sparse=False)
epochs = 20
functions.np.random.seed(0)
print(Y[0:10].compute())
encY = le.fit_transform(Y.ravel())
dummy_y = enc.fit_transform(encY.reshape(-1, 1))
print(dummy_y)
print('scaling X')
if config.scaleX == True:
def plot_the_Graph(self,graph,filename):
nx.draw(graph)
write_path=self.basic_path[5]
plt.save_fig(write_path+filename)
def save_fig(fig_id, tight_layout=True, fig_extension="png", resolution=300):
path = os.path.join(IMGS_DIR, fig_id + "." + fig_extension)
print("Salving figure", fig_id)
if tight_layout:
plt.tight_layout()
plt.save_fig(path, format=fig_extension, dpi=resolution)
def plot_video_timelines(
video_rows,
show_segment_boundaries=False,
segment_color_map={
'commercial': 'LightGray', 'non-commercial': 'LightCoral'
},
interval_label_color_map={},
discrete_label_shape_map={},
max_length=None, # timedelta for video length
out_file=None, # save to a file (e.g., a pdf)
show_legend=True,
min_y_margin=None
):
fig = plt.figure()
fig.set_size_inches(14, 2 * len(video_rows))
ax = fig.add_subplot(111)
# Plot grid for x axis
ax.grid(color='Grey', which='major', linestyle='-', linewidth=1, axis='x')
ax.grid(color='LightGrey', which='minor', linestyle='--', linewidth=1, axis='x')
# Plot grid separating the shows on the y axis
ax.grid(color='LightGrey', which='minor', linestyle='--', linewidth=1, axis='y')
# Do not add labels more than once so they appear only once in the
# legend
defined_labels = set()
y_labels = []
for i, row in enumerate(video_rows):
y_labels.append('[{} {}]\n{}\n(id: {})'.format(row.channel, row.date, row.show, row.id))
# Plot segments
for segment_info in row.segments:
height = (0.05 + 0.95 * segment_info.display_value) / 2.1 # Use a little less
# than half of the gap
width = segment_info.end_time.total_seconds() - segment_info.start_time.total_seconds()
rect = Rectangle(
(
segment_info.start_time.total_seconds(),
i
),
width,
height,
fc=segment_color_map[segment_info.display_label],
label=segment_info.display_label if segment_info.display_label not in defined_labels else None
)
defined_labels.add(segment_info.display_label)
ax.add_patch(rect)
# Plot shot boundaries
if show_segment_boundaries:
for segment_info in row.segments.values():
segment_end = segment_info.end_time.total_seconds()
plt.plot(
(segment_end, segment_end),
(i, i + 0.25),
'Black',
linewidth=1
)
vert_ofs = 0.05
# Plot interval labels
for label, intervals in row.interval_labels.items():
for interval in intervals:
plt.plot(
[td.total_seconds() for td in interval],
(i - vert_ofs, i- vert_ofs),
interval_label_color_map[label],
linewidth=2,
label=label if label not in defined_labels else None
)
defined_labels.add(label)
vert_ofs += 0.05
# Plot discrete labels
for label, offsets in row.discrete_labels.items():
plt.scatter(
[td.total_seconds() for td in offsets],
[i - vert_ofs] * len(offsets),
s=2,
c='Black',
marker=discrete_label_shape_map[label],
label=label if label not in defined_labels else None
)
defined_labels.add(label)
vert_ofs += 0.05
ax.set_yticks([0.5 + i for i in range(len(video_rows))], minor=True)
# TODO: this is a hack to fix the margins
if min_y_margin is not None:
ax.text(-min_y_margin, 0, '.')
plt.yticks(range(len(video_rows)), y_labels)
plt.ylim([-0.5, len(video_rows) - 0.5])
if max_length is None:
max_length = max([x.length.total_seconds() for x in video_rows])
else:
max_length = max_length.total_seconds()
plt.xlim([0, int(max_length)])
# Minor ticks every 15 min
ax.set_xticks(range(0, int(max_length), 900), minor=True)
# Major ticks every hour
plt.xticks(range(0, int(max_length), 3600))
plt.xlabel('video time (s)')
if show_legend:
plt.legend(loc=0, frameon=True, edgecolor='Black', facecolor='White', framealpha=0.7)
if out_file is not None:
plt.save_fig(out_file)
plt.show()
def plot_bird_label_scores(automated_df, human_df, save_fig=False):
"""
Function to visualize automated and human annotation scores across an audio
clip.
Args:
automated_df (Dataframe)
- Dataframe of automated labels for one clip
human_df (Dataframe)
- Dataframe of human labels for one clip.
plot_fig (boolean)
- Whether or not the efficiency statistics should be displayed.
save_fig (boolean)
- Whether or not the plot should be saved within a file.
Returns:
Dataframe with statistics comparing the automated and human labeling.
"""
duration = automated_df["CLIP LENGTH"].to_list()[0]
SAMPLE_RATE = automated_df["SAMPLE RATE"].to_list()[0]
# Initializing two arrays that will represent the
# labels with respect to the audio clip
# print(SIGNAL.shape)
human_arr = np.zeros((int(SAMPLE_RATE * duration), ))
bot_arr = np.zeros((int(SAMPLE_RATE * duration), ))
folder_name = automated_df["FOLDER"].to_list()[0]
clip_name = automated_df["IN FILE"].to_list()[0]
# Placing 1s wherever the au
for row in automated_df.index:
minval = int(round(automated_df["OFFSET"][row] * SAMPLE_RATE, 0))
maxval = int(
round(
(automated_df["OFFSET"][row] + automated_df["DURATION"][row]) *
SAMPLE_RATE, 0))
bot_arr[minval:maxval] = 1
for row in human_df.index:
minval = int(round(human_df["OFFSET"][row] * SAMPLE_RATE, 0))
maxval = int(
round((human_df["OFFSET"][row] + human_df["DURATION"][row]) *
SAMPLE_RATE, 0))
human_arr[minval:maxval] = 1
human_arr_flipped = 1 - human_arr
bot_arr_flipped = 1 - bot_arr
true_positive_arr = human_arr * bot_arr
false_negative_arr = human_arr * bot_arr_flipped
false_positive_arr = human_arr_flipped * bot_arr
true_negative_arr = human_arr_flipped * bot_arr_flipped
IoU_arr = human_arr + bot_arr
IoU_arr[IoU_arr == 2] = 1
plt.figure(figsize=(22, 10))
plt.subplot(7, 1, 1)
plt.plot(human_arr)
plt.title("Ground Truth for " + clip_name)
plt.subplot(7, 1, 2)
plt.plot(bot_arr)
plt.title("Automated Label for " + clip_name)
# Visualizing True Positives for the Automated Labeling
plt.subplot(7, 1, 3)
plt.plot(true_positive_arr)
plt.title("True Positive for " + clip_name)
# Visualizing False Negatives for the Automated Labeling
plt.subplot(7, 1, 4)
plt.plot(false_negative_arr)
plt.title("False Negative for " + clip_name)
plt.subplot(7, 1, 5)
plt.plot(false_positive_arr)
plt.title("False Positive for " + clip_name)
plt.subplot(7, 1, 6)
plt.plot(true_negative_arr)
plt.title("True Negative for " + clip_name)
plt.subplot(7, 1, 7)
plt.plot(IoU_arr)
plt.title("Union for " + clip_name)
plt.tight_layout()
if save_fig:
x = clip_name.split(".")
clip_name = x[0]
plt.save_fig(clip_name + "_label_plot.png")
def plot(matrix, weights=None, title="Prediction Matrix", save_plot=False):
if save_plot and not os.path.isdir("plots"): os.makedirs("plots")
if len(matrix[0]) == 3: # if 1D inputs, excluding bias and ys
fig, ax = plt.subplots()
ax.set_title(title)
ax.set_xlabel("i1")
ax.set_ylabel("Classifications")
if weights != None:
y_min = -0.1
y_max = 1.1
x_min = 0.0
x_max = 1.1
y_res = 0.001
x_res = 0.001
ys = np.arange(y_min, y_max, y_res)
xs = np.arange(x_min, x_max, x_res)
zs = []
for cur_y in np.arange(y_min, y_max, y_res):
for cur_x in np.arange(x_min, x_max, x_res):
zs.append(predict([1.0, cur_x], weights))
xs, ys = np.meshgrid(xs, ys)
zs = np.array(zs)
zs = zs.reshape(xs.shape)
cp = plt.contourf(xs,
ys,
zs,
levels=[-1, -0.0001, 0, 1],
colors=('b', 'r'),
alpha=0.1)
c1_data = [[], []]
c0_data = [[], []]
for i in range(len(matrix)):
cur_i1 = matrix[i][1]
cur_y = matrix[i][-1]
if cur_y == 1:
c1_data[0].append(cur_i1)
c1_data[1].append(1.0)
else:
c0_data[0].append(cur_i1)
c0_data[1].append(0.0)
plt.xticks(np.arange(x_min, x_max, 0.1))
plt.yticks(np.arange(y_min, y_max, 0.1))
plt.xlim(0, 1.05)
plt.ylim(-0.05, 1.05)
c0s = plt.scatter(c0_data[0],
c0_data[1],
s=40.0,
c='r',
label='Class -1')
c1s = plt.scatter(c1_data[0],
c1_data[1],
s=40.0,
c='b',
label='Class 1')
plt.legend(fontsize=10, loc=1)
if save_plot:
plt.save_fig(os.path.join("plots", title + ".png"),
bbox_inches='tight',
dpi=200)
else:
plt.show()
return
if len(matrix[0]) == 4: # if 2D inputs, excluding bias and ys
fig, ax = plt.subplots()
ax.set_title(title)
ax.set_xlabel("i1")
ax.set_ylabel("i2")
if weights != None:
map_min = 0.0
map_max = 1.1
y_res = 0.001
x_res = 0.001
ys = np.arange(map_min, map_max, y_res)
xs = np.arange(map_min, map_max, x_res)
zs = []
for cur_y in np.arange(map_min, map_max, y_res):
for cur_x in np.arange(map_min, map_max, x_res):
zs.append(predict([1.0, cur_x, cur_y], weights))
xs, ys = np.meshgrid(xs, ys)
zs = np.array(zs)
zs = zs.reshape(xs.shape)
cp = plt.contourf(xs,
ys,
zs,
levels=[-1, -0.0001, 0, 1],
colors=('b', 'r'),
alpha=0.1)
c1_data = [[], []]
c0_data = [[], []]
for i in range(len(matrix)):
cur_i1 = matrix[i][1]
cur_i2 = matrix[i][2]
cur_y = matrix[i][-1]
if cur_y == 1:
c1_data[0].append(cur_i1)
c1_data[1].append(cur_i2)
else:
c0_data[0].append(cur_i1)
c0_data[1].append(cur_i2)
plt.xticks(np.arange(0.0, 1.1, 0.1))
plt.yticks(np.arange(0.0, 1.1, 0.1))
plt.xlim(0, 1.05)
plt.ylim(0, 1.05)
c0s = plt.scatter(c0_data[0],
c0_data[1],
s=40.0,
c='r',
label='Class -1')
c1s = plt.scatter(c1_data[0],
c1_data[1],
s=40.0,
c='b',
label='Class 1')
plt.legend(fontsize=10, loc=1)
if save_plot:
plt.save_fig(os.path.join("plots", title + ".png"),
bbox_inches='tight',
dpi=200)
else:
plt.show()
return
print("Matrix dimensions not covered.")
nonzeros = float(frame[frame >= threshold].size)
listy.append(zeros / nonzeros)
times.append(frame_number)
if (zeros / nonzeros) <= proportion_dark:
newfmf.add_frame(frame, timestamp)
sumframe = sumframe + frame
listynew.append(zeros / nonzeros)
timesnew.append(frame_number)
else:
pass
newfmf.close()
#plot proportion of black pixels before and after
plt.plot(times, listy, color='b')
plt.plot(timesnew, listynew, color='r')
plt.save_fig(sys.argv[2] + '_zeros.svg')
plt.show()
#plot average intensity (vignette)
vignette = sumframe / len(timesnew)
plt.imshow(vignette)
plt.save_fig(sys.argv[2] + '_vignette.svg')
plt.show()
#Save vignette values
DataFrame(vignette).to_csv(sys.argv[2] + '_array.csv', sep=',')
print 'done.'
train_sample = [resize(i, input_shape) for i in train_sample]
train_sample = np.array(train_sample)
train_labeled_numeric = [
test_right_generator.class_indices[j] for j in train_label
]
test_l = test_right_generator.n
counts = 0
acc = []
try:
model.save_weight('siamese.h5')
except:
pass
for (i, j) in test_right_generator:
test_tmp = np.zeros_like(train_sample)
test_tmp[:] = i[0]
scores = model.predict([train_sample, test_tmp])
pred_y = train_labeled_numeric[scores.argmax()]
acc.append(pred_y == int(j[0]))
counts += 1
print(counts)
if counts >= test_right_generator.n:
break
import pandas as pd
h = pd.DataFrame(hist.history)
plt.figure()
h.plot()
plt.save_fig('siamese.png')
acc = sum(acc) / len(acc)
print(acc)
else:
select_id = None
if mc_info[event_no][2] < 60:
continue
print(
"Mc Info: [event_id, particle_type, energy, isCC, bjorkeny, dir_x/y/z, time]\n",
mc_info[event_no])
#event: (1,11,13,18,31)
#prediction: (1,11,13,18,3)
prediction = encoder.predict(event.reshape((1, ) + event.shape))
prediction = np.reshape(prediction,
prediction.shape[1:]) #11,13,18,3
xyz_event = np.add(event, xs_mean)
xyz_event = np.sum(event.reshape(data.shape[1:]),
axis=-1) #11,13,18
titles = [
"Summed over channel id", "Neuron 1", "Neuron 2", "Neuron 3"
]
fig = make_4_plots(xyz_event, prediction[:, :, :, 0],
prediction[:, :, :, 1], prediction[:, :, :, 2],
n_bins[:-1], titles)
plt.show(fig)
plt.save_fig("fig" + str(event_id) + ".pdf")
else:
print("ploty type unknow")
raise ()
def save(f):
plt.save_fig(f)
def plot_and_save(dataframe, save_as):
plot_combined(dataframe)
plt.save_fig(str(save_as), bbox_inches='tight')
plt.clf()
# Import numpy and matplotlib.pyplot
import numpy as np
import matplotlib.pyplot as plt
# Generate two 1-D arrays: u, v
u = np.linspace(-2, 2, 41)
v = np.linspace(-1, 1, 21)
# Generate 2-D arrays from u and v: X, Y
X, Y = np.meshgrid(u, v)
# Compute Z based on X and Y
Z = np.sin(3 * np.sqrt(X**2 + Y**2))
# Display the resulting image with pcolor()
plt.pcolor(Z)
plt.show()
# Save the figure to 'sine_mesh.png'
plt.save_fig('sine_mesh.png')
if args.decay_every == -1:
args.decay_every = np.inf
if args.train:
demo = False
get = False
else:
demo = True
get = True
args.add_epochs = 0
im = began_train(start_epoch=args.start_epoch,
add_epochs=args.add_epochs,
batch_size=args.batch_size,
hidden_size=args.hidden_size,
gpu_id=args.gpuid,
demo=demo,
get=get,
save_every=args.save_every,
decay_every=args.decay_every,
batch_norm=args.batch_norm,
start_learn_rate=args.start_learn_rate)
if not args.train:
import matplotlib.pyplot as plt
for n in range(8):
im_to_save = im[n].reshape([800, 3])
plt.plot(im_to_save)
plt.save_fig(args.outdir + '/out_{}.jpg'.format(n))