-
Notifications
You must be signed in to change notification settings - Fork 13
/
Copy pathdetection_manager.py
300 lines (264 loc) · 16.1 KB
/
detection_manager.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
import sys
import subprocess
import time
import numpy as np
import os
import cv2 as cv
import operator
import math
import torch
from PIL import Image
from torch.autograd import Variable
import torch.nn.functional as F
sys.path.append('../')
from simple_trainer_manager import SimpleTrainer
from metrics.metrics import compute_accuracy, compute_confusion_matrix, extract_stats_from_confm, compute_mIoU
from utils.plot import Compute_plot
from metrics.object_detection import Compute_kitti_AP
class Detection_Manager(SimpleTrainer):
def __init__(self, cf, model):
super(Detection_Manager, self).__init__(cf, model)
class train(SimpleTrainer.train):
def __init__(self, logger_stats, model, cf, validator, stats, msg):
super(Detection_Manager.train, self).__init__(logger_stats, model, cf, validator, stats, msg)
if self.cf.resume_experiment:
self.msg.msg_stats_best = 'Best case [%s]: epoch = %d, mIoU = %.2f, acc= %.2f, loss = %.5f\n' % (
self.cf.save_condition, self.model.net.best_stats.epoch, 100 * self.model.net.best_stats.val.mIoU,
100 * self.model.net.best_stats.val.acc, self.model.net.best_stats.val.loss)
def training_loop(self, epoch, train_loader, epoch_bar):
# Train epoch
for i, data in enumerate(train_loader):
# Read Data
inputs, loc_targets, cls_targets = data
N, w, h, c = inputs.size()
self.inputs = Variable(inputs).cuda()
self.loc_targets = Variable(loc_targets).cuda()
self.cls_targets = Variable(cls_targets).cuda()
# Predict model
self.model.optimizer.zero_grad()
self.loc_preds, self.cls_preds = self.model.net(self.inputs)
# Compute gradients
self.compute_gradients()
# Compute batch stats
self.train_loss.update(float(self.loss[0].cpu().data[0]), N)
# confm = compute_confusion_matrix(predictions, self.labels.cpu().data.numpy(), self.cf.num_classes,
# self.cf.void_class)
# self.confm_list = map(operator.add, self.confm_list, confm)
if self.cf.normalize_loss:
self.stats.train.loss = self.train_loss.avg
else:
self.stats.train.loss = self.train_loss.avg
if not self.cf.debug:
# Save stats
self.save_stats_batch((epoch - 1) * self.train_num_batches + i)
# Update epoch messages
self.update_epoch_messages(epoch_bar, self.global_bar, self.train_num_batches, epoch, i)
# Save model without training
if self.cf.epochs == 0:
self.model.save_model()
def validate_epoch(self, valid_set, valid_loader, early_Stopping, epoch, global_bar):
if valid_set is not None and valid_loader is not None:
# Set model in validation mode
self.model.net.eval()
self.model.net.training = False
self.validator.start(valid_set, valid_loader, 'Epoch Validation', epoch,
global_bar=global_bar, save_folder=self.cf.temp_folder)
# Early stopping checking
if self.cf.early_stopping:
early_Stopping.check(self.stats.train.loss, self.stats.val.AP)
if early_Stopping.stop == True:
self.stop = True
# Set model in training mode
self.model.net.training = True
self.model.net.train()
def compute_gradients(self):
self.loss = self.model.loss(self.loc_preds, self.loc_targets, self.cls_preds, self.cls_targets)
self.loss[0].backward()
self.model.optimizer.step()
def update_messages(self, epoch, epoch_time, new_best):
# Update logger
epoch_time = time.time() - epoch_time
self.logger_stats.write('\t Epoch step finished: %ds \n' % (epoch_time))
# Compute best stats
self.msg.msg_stats_last = '\nLast epoch: loss = %.5f, mean_AP = %.2f' % (self.stats.val.loss, self.stats.val.AP)
for label in range(len(self.cf.labels)):
self.msg.msg_stats_last += ', AP_%s %.2f' % (self.cf.labels[label], self.stats.val.AP_perclass[label])
if new_best:
self.msg.msg_stats_best = '\n Best case [%s]: epoch = %d, avg_loss = %.5f' % (
self.cf.save_condition,epoch, self.stats.val.loss)
for label in range(len(self.cf.labels)):
self.msg.msg_stats_best += ', AP_%s %.2f' % (
self.cf.labels[label], self.stats.val.AP_perclass[label])
# msg_confm = self.stats.val.get_confm_str()
# self.logger_stats.write(msg_confm)
# self.msg.msg_stats_best = self.msg.msg_stats_best + '\nConfusion matrix:\n' + msg_confm
def update_epoch_messages(self, epoch_bar, global_bar, train_num_batches, epoch, batch):
# Update progress bar
epoch_bar.set_msg('loc_loss = %.5f, cls_loss = %.5f, batch_loss = %.5f, avg_loss = %.5f' %
(self.loss[1], self.loss[2], float(self.loss[0].cpu().data[0]), self.stats.train.loss))
self.msg.last_str = epoch_bar.get_message(step=True)
global_bar.set_msg(self.msg.accum_str + self.msg.last_str + self.msg.msg_stats_last + \
self.msg.msg_stats_best)
global_bar.update()
# writer.add_scalar('train_loss', train_loss.avg, curr_iter)
# Display progress
curr_iter = (epoch - 1) * train_num_batches + batch + 1
if (batch + 1) % math.ceil(train_num_batches / 20.) == 0:
self.logger_stats.write('[Global iteration %d], [iter %d / %d], [train loss %.5f] \n' % (
curr_iter, batch + 1, train_num_batches, self.stats.train.loss))
def compute_stats(self, confm_list, train_loss):
# TP_list, TN_list, FP_list, FN_list = extract_stats_from_confm(confm_list)
# mean_IoU = compute_mIoU(TP_list, FP_list, FN_list)
# mean_accuracy = compute_accuracy_segmentation(TP_list, FN_list)
# self.stats.train.acc = np.nanmean(mean_accuracy)
# self.stats.train.mIoU_perclass = mean_IoU
# self.stats.train.mIoU = np.nanmean(mean_IoU)
if train_loss is not None:
self.stats.val.loss = train_loss.avg
def save_stats_epoch(self, epoch):
# Save logger
if epoch is not None:
# Epoch loss tensorboard
self.writer.add_scalar('losses/epoch', self.stats.train.loss, epoch)
self.writer.add_scalar('metrics/accuracy', 100.*self.stats.train.acc, epoch)
self.writer.add_scalar('metrics/mIoU', 100.*self.stats.train.mIoU, epoch)
# conf_mat_img = confm_metrics2image(self.stats.train.get_confm_norm(), self.cf.labels)
# self.writer.add_image('metrics/conf_matrix', conf_mat_img, epoch)
class validation(SimpleTrainer.validation):
def __init__(self, logger_stats, model, cf, stats, msg):
super(Detection_Manager.validation, self).__init__(logger_stats, model, cf, stats, msg)
def validation_loop(self, epoch, valid_loader, valid_set, bar, global_bar, save_folder=None):
if epoch is None:
if not os.path.exists(os.path.join(self.cf.predict_path_output,'data')):
os.makedirs(os.path.join(self.cf.predict_path_output,'data'))
else:
if not os.path.exists(os.path.join(save_folder,'data')):
os.makedirs(os.path.join(save_folder,'data'))
val_epoch_images = open(os.path.join(save_folder, "val_epoch_images.txt"), 'w')
for vi, data in enumerate(valid_loader):
# Read data
inputs, loc_targets, cls_targets, gt_path, size = data
n_images, _, _, _ = inputs.size()
img_name = gt_path[0].split('/')[-1].split('.')[0]
inputs = Variable(inputs).cuda()
loc_targets = Variable(loc_targets).cuda()
cls_targets = Variable(cls_targets).cuda()
if epoch is None:
f = open(os.path.join(self.cf.predict_path_output, 'data', img_name + ".txt"), 'w')
else:
f = open(os.path.join(save_folder , 'data', img_name + ".txt"), 'w')
# Predict model
with torch.no_grad():
loc_preds, cls_preds = self.model.net(inputs)
# self.val_loss.update(float(self.model.loss(loc_preds, loc_targets, cls_preds,
# cls_targets).cpu().data[0] / n_images), n_images)
box_preds, label_preds, score_preds = self.model.box_coder.decode(
loc_preds.cpu().data.squeeze(),
F.softmax(cls_preds.squeeze(), dim=1).cpu().data,
score_thresh=0.25)
if len(box_preds) > 0:
box_preds = box_preds.cpu().numpy()
label_preds = label_preds.cpu().numpy()
score_preds = score_preds.cpu().numpy()
# print(len(box_preds))
# print(label_preds.size)
# print(score_preds.size)
for det in range(len(box_preds)):
if score_preds[det] >= 0:
box_preds[det] = box_preds[det] / [size[0], size[1], size[0], size[1]]
f.write(self.cf.labels[label_preds[det]] + ' ' + str(0) + ' ' + str(0) + ' ' + '-10' + ' ' \
+ str(box_preds[det][0]) + ' ' + str(box_preds[det][1]) + ' ' \
+ str(box_preds[det][2]) + ' ' + str(box_preds[det][3]) + ' ' \
+ '-1 -1 -1 -1000 -1000 -1000 ' \
+ '-1000 ' \
+ str(score_preds[det]) + '\n'
)
f.close()
if epoch is not None:
val_epoch_images.write(gt_path[0] + '\n')
# Compute batch stats
# self.val_loss.update(float(self.model.loss(outputs, gts).cpu().data[0] / n_images), n_images)
# confm = compute_confusion_matrix(predictions, gts.cpu().data.numpy(), self.cf.num_classes,
# self.cf.void_class)
# confm_list = map(operator.add, confm_list, confm)
# Save epoch stats
# self.stats.val.conf_m = confm_list
if not self.cf.normalize_loss:
self.stats.val.loss = self.val_loss.avg
else:
self.stats.val.loss = self.val_loss.avg
# Save predictions and generate overlaping
# self.update_tensorboard(inputs.cpu(), gts.cpu(),
# predictions, epoch, range(vi * self.cf.valid_batch_size,
# vi * self.cf.valid_batch_size +
# np.shape(predictions)[0]),
# valid_set.num_images)
# Update messages
self.update_msg(bar, global_bar)
# Calculate stats for detection
# extern kitti evaluation call
if epoch is not None:
val_epoch_images.close()
evaluation_bash_comand = "./devkit_kitti_txt/cpp/evaluate_object_txt %s %s %s" % (
self.cf.temp_folder, os.path.join(self.cf.temp_folder,"val_epoch_images.txt"), self.cf.temp_folder)
else:
evaluation_bash_comand = "./devkit_kitti_txt/cpp/evaluate_object_txt %s %s %s" % (
self.cf.predict_path_output, self.cf.valid_gt_txt, self.cf.predict_path_output)
process = subprocess.Popen(evaluation_bash_comand, shell=True)
process.communicate()
def compute_stats(self, confm_list, val_loss):
for label in self.cf.labels:
scores = Compute_kitti_AP(os.path.join(self.cf.temp_folder,"stats_%s_detection.txt" %(label.lower())))
# print scores
self.stats.val.AP_perclass.append(scores[1])
self.stats.val.AP = np.mean(np.asarray(self.stats.val.AP_perclass, dtype=np.float32))
if val_loss is not None:
self.stats.val.loss = val_loss.avg
def save_stats(self, epoch):
# Save logger
if epoch is not None:
# add log
self.logger_stats.write('----------------- Epoch scores summary ------------------------- \n')
self.logger_stats.write('[epoch %d], [val loss %.5f]' % (
epoch, self.stats.val.loss))
for label in range(len(self.cf.labels)):
self.logger_stats.write(', [AP_%s %.2f]' % (self.cf.labels[label], self.stats.val.AP_perclass[label]))
# add scores to tensorboard
self.writer.add_scalar('metrics/AP_%s'%(self.cf.labels[label]),
100.*self.stats.val.AP_perclass[label], epoch)
self.logger_stats.write('\n---------------------------------------------------------------- \n')
# add scores to tensorboard
self.writer.add_scalar('losses/epoch', self.stats.val.loss, epoch)
else:
self.logger_stats.write('----------------- Scores summary -------------------- \n')
self.logger_stats.write('[val loss %.5f]' % (
self.stats.val.loss))
for label in range(len(self.cf.labels)):
self.logger_stats.write(', [AP_%s %.2f]' % (self.cf.labels[label], self.stats.val.AP_perclass[label]))
self.logger_stats.write('\n---------------------------------------------------------------- \n')
def update_msg(self, bar, global_bar):
# self.compute_stats(np.asarray(self.stats.val.conf_m), None)
# bar.set_msg(', mIoU: %.02f' % (100.*np.nanmean(self.stats.val.mIoU)))
if global_bar==None:
# Update progress bar
bar.update()
else:
self.msg.eval_str = '\n' + bar.get_message(step=True)
global_bar.set_msg(self.msg.accum_str + self.msg.last_str + self.msg.msg_stats_last + self.msg.msg_stats_best + self.msg.eval_str)
global_bar.update()
def update_tensorboard(self,inputs,gts,predictions,epoch,indexes,val_len):
if epoch is not None and self.cf.color_map is not None:
save_img(self.writer, inputs, gts, predictions, epoch, indexes, self.cf.predict_to_save, val_len,
self.cf.color_map, self.cf.labels, self.cf.void_class, n_legend_rows=3)
class predict(SimpleTrainer.predict):
def __init__(self, logger_stats, model, cf):
super(Detection_Manager.predict, self).__init__(logger_stats, model, cf)
def write_results(self, predictions, img_name, img_shape):
path = os.path.join(self.cf.predict_path_output, img_name[0])
predictions = predictions[0]
predictions = Image.fromarray(predictions.astype(np.uint8))
if self.cf.resize_image_test is not None:
predictions = predictions.resize((self.cf.original_size[1],
self.cf.original_size[0]), resample=Image.BILINEAR)
predictions = np.array(predictions)
cv.imwrite(path, predictions)