def data_aug(train_path,
num_rois,
hf=True,
vf=False,
rot_90=False,
num_epoches=2000,
output_weight_path='weights/{}faster_rcnn.hdf5'.format(
time.localtime(time.time()))):
C = config.Config()
C.num_rois = int(num_rois)
C.use_horizontal_flips = hf
C.use_vertical_flips = vf
C.rot_90 = rot_90
C.model_path = output_weight_path
all_imgs, class_count, class_mapping = get_data(train_path)
for img_data in all_imgs:
img_aug, img = data_augment.augment(img_data, C, augment=True)
if img_aug['filepath'] != img_data['filepath']:
all_imgs.append(img_aug)
else:
pass
import math
import os.path
from model import config
C = config.Config()
def anchors():
return mk_anchors(C.TRAIN_SCALES[0], C.TRAIN_SCALES[1], C.SIZE_PATTERN,
C.STEP_PATTERN)
def reglayer_scale(size, num_layer, size_the):
reg_layer_size = []
for i in range(num_layer + 1):
size = math.ceil(size / 2.)
if i >= 2:
reg_layer_size += [size]
if i == num_layer and size_the != 0:
reg_layer_size += [size - size_the]
return reg_layer_size
def get_scales(size, size_pattern):
size_list = []
for x in size_pattern:
size_list += [round(x * size, 2)]
return size_list
def aspect_ratio(num):
transform_test)
mdl = model.build_extractor_fpn(cfg)
mdl.load_state_dict(torch.load(mdl_path))
if torch.cuda.is_available():
print "GPU is available, so the model will be trained on GPU"
mdl.cuda()
mdl.eval()
total_test_step = len(test_loader)
#TODO: could implmement 10 crops evaluation
for i, (images, labels) in enumerate(test_loader):
print 'Step [%d/%d]'.format(i, total_test_step)
res, _ = mdl(images)
pre = res[0]
pred = torch.cat((pred, pre), 0)
gt = torch.cat((gt, labels), 0)
AUROCs = tools.compute_AUCs(gt, pred)
AUROC_avg = np.array(AUROCs).mean()
print('The average AUROC is {AUROC_avg:.3f}'.format(AUROC_avg=AUROC_avg))
for i in range(14):
print('The AUROC of {} is {}'.format(CLASS_NAMES[i], AUROCs[i]))
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--mdl_path',
type=str,
default='./checkpoints/',
help='path for trained encoder')
cfg = config.Config()
args = parser.parse_args()
test(cfg, args)
def main():
args = parser.parse_args()
time_stamp = "{0:%Y%m%d-%H%M%S}".format(datetime.now())
save_name = os.path.join(args.save_dir, "train_{}".format(time_stamp))
if not(os.path.isdir(args.save_dir)):
os.makedirs(args.save_dir)
if args.path == None:
raise OSError("path to annotation file must be required.")
C = config.Config()
C.config_filename = save_name + "_config.pickle"
C.model_path = save_name + "_model.hdf5"
C.use_horizontal_flips = bool(args.horizontal_flips)
C.use_vertical_flips = bool(args.vertical_flips)
C.rot_90 = bool(args.rot_90)
all_imgs, classes_count, class_mapping = get_data(args.path)
C.class_mapping = class_mapping
with open(C.config_filename, 'wb') as config_f:
pickle.dump(C,config_f)
print("-------------------------------")
print('path to config file : {}'.format(C.config_filename))
print("-------------------------------")
train_imgs = [s for s in all_imgs if s['imageset'] == 'trainval']
val_imgs = [s for s in all_imgs if s['imageset'] == 'test']
data_gen_train = data_generators.get_anchor_gt(train_imgs, classes_count, C, K.image_dim_ordering(), mode='train')
data_gen_val = data_generators.get_anchor_gt(val_imgs, classes_count, C, K.image_dim_ordering(), mode='val')
model_rpn, model_classifier, model_all = faster_rcnn.get_model(C, classes_count)
losses = np.zeros((args.n_iters, 5))
rpn_accuracy_rpn_monitor, rpn_accuracy_for_epoch = [], []
best_loss = np.Inf
with open('out.csv', 'w') as f:
f.write('Accuracy,RPN classifier,RPN regression,Detector classifier,Detector regression,Total')
f.write('\t')
iter_num = 0
t0 = start_time = time.time()
try:
for epoch_num in range(args.n_epochs):
progbar = generic_utils.Progbar(args.n_iters)
print('Epoch {}/{}'.format(epoch_num + 1, args.n_epochs))
while True:
try:
if len(rpn_accuracy_rpn_monitor) == args.n_iters and C.verbose:
mean_overlapping_bboxes = float(sum(rpn_accuracy_rpn_monitor))/len(rpn_accuracy_rpn_monitor)
rpn_accuracy_rpn_monitor = []
print('Average number of overlapping bounding boxes from RPN = {} for {} previous iterations'.format(mean_overlapping_bboxes, args.n_iters))
if mean_overlapping_bboxes == 0:
print('RPN is not producing bounding boxes that overlap the ground truth boxes. Check RPN settings or keep training.')
X, Y, img_data = next(data_gen_train)
loss_rpn = model_rpn.train_on_batch(X, Y)
P_rpn = model_rpn.predict_on_batch(X)
R = roi_helpers.rpn_to_roi(P_rpn[0], P_rpn[1], C, K.image_dim_ordering(), use_regr=True, overlap_thresh=0.7, max_boxes=300)
# note: calc_iou converts from (x1,y1,x2,y2) to (x,y,w,h) format
X2, Y1, Y2 = roi_helpers.calc_iou(R, img_data, C, class_mapping)
neg_samples = np.where(Y1[0, :, -1] == 1)
pos_samples = np.where(Y1[0, :, -1] == 0)
if len(neg_samples) > 0:
neg_samples = neg_samples[0]
else:
neg_samples = []
if len(pos_samples) > 0:
pos_samples = pos_samples[0]
else:
pos_samples = []
rpn_accuracy_rpn_monitor.append(len(pos_samples))
rpn_accuracy_for_epoch.append((len(pos_samples)))
if len(pos_samples) < C.num_rois//2:
selected_pos_samples = pos_samples.tolist()
else:
selected_pos_samples = np.random.choice(pos_samples, C.num_rois//2, replace=False).tolist()
try:
selected_neg_samples = np.random.choice(neg_samples, C.num_rois - len(selected_pos_samples), replace=False).tolist()
except:
selected_neg_samples = np.random.choice(neg_samples, C.num_rois - len(selected_pos_samples), replace=True).tolist()
sel_samples = selected_pos_samples + selected_neg_samples
loss_class = model_classifier.train_on_batch([X, X2[:, sel_samples, :]], [Y1[:, sel_samples, :], Y2[:, sel_samples, :]])
if iter_num == args.n_iters:
loss_rpn_cls = np.mean(losses[:, 0])
loss_rpn_regr = np.mean(losses[:, 1])
loss_class_cls = np.mean(losses[:, 2])
loss_class_regr = np.mean(losses[:, 3])
class_acc = np.mean(losses[:, 4])
mean_overlapping_bboxes = float(sum(rpn_accuracy_for_epoch)) / len(rpn_accuracy_for_epoch)
rpn_accuracy_for_epoch = []
if C.verbose:
print('Mean number of bounding boxes from RPN overlapping ground truth boxes: {}'.format(mean_overlapping_bboxes))
print('Classifier accuracy for bounding boxes from RPN: {}'.format(class_acc))
print('Loss RPN classifier: {}'.format(loss_rpn_cls))
print('Loss RPN regression: {}'.format(loss_rpn_regr))
print('Loss Detector classifier: {}'.format(loss_class_cls))
print('Loss Detector regression: {}'.format(loss_class_regr))
print('Elapsed time: {}[s]'.format(time.time() - start_time))
target_text_file = open('out.csv', 'a')
target_text_file.write('{},{},{},{},{},{}'.format(class_acc, loss_rpn_cls,
loss_rpn_regr, loss_class_cls, loss_class_regr,
loss_rpn_cls + loss_rpn_regr + loss_class_cls + loss_class_regr))
target_text_file.write('\t')
curr_loss = loss_rpn_cls + loss_rpn_regr + loss_class_cls + loss_class_regr
iter_num = 0
start_time = time.time()
if curr_loss < best_loss:
if C.verbose:
print('Total loss decreased from {} to {}, saving weights'.format(best_loss,curr_loss))
best_loss = curr_loss
model_all.save_weights(C.model_path)
break
losses[iter_num, 0] = loss_rpn[1]
losses[iter_num, 1] = loss_rpn[2]
losses[iter_num, 2] = loss_class[1]
losses[iter_num, 3] = loss_class[2]
losses[iter_num, 4] = loss_class[3]
iter_num += 1
progbar.update(iter_num, [('rpn_cls', np.mean(losses[:iter_num, 0])), ('rpn_regr', np.mean(losses[:iter_num, 1])),
('detector_cls', np.mean(losses[:iter_num, 2])), ('detector_regr', np.mean(losses[:iter_num, 3]))])
except Exception as e:
print('Exception: {}'.format(e))
continue
except KeyboardInterrupt:
t1 = time.time()
print('\nIt took {:.2f}s'.format(t1-t0))
sys.exit('Keyboard Interrupt')
print("training is done")
print("-------------------------------")
print('path to config file : {}'.format(C.config_filename))
print("-------------------------------")
def main(train_path,
num_rois,
hf=True,
vf=False,
rot_90=False,
num_epoches=2000,
output_weight_path='weights/{}faster_rcnn.hdf5'.format(
time.localtime(time.time()))):
C = config.Config()
C.num_rois = int(num_rois)
C.use_horizontal_flips = hf
C.use_vertical_flips = vf
C.rot_90 = rot_90
C.model_path = output_weight_path
all_imgs, class_count, class_mapping = get_data(train_path)
if 'bg' not in class_count:
class_count['bg'] = 0
class_mapping['bg'] = len(class_mapping)
C.class_mapping = class_mapping
inv_map = {v: k for k, v in class_mapping.items()}
print('classes_count:', class_count)
print('class_mapping', class_mapping)
train_imgs = [s for s in all_imgs if s['type'] == 'train']
val_imgs = [s for s in all_imgs if s['type'] == 'val']
test_imgs = [s for s in all_imgs if s['type'] == 'test']
print("the number of train", len(train_imgs))
print('the number of val', len(val_imgs))
print('the number of test', len(test_imgs))
data_gen_train = data_generators.get_anchor_gt(train_imgs,
class_count,
C,
K.image_dim_ordering(),
mode='train')
data_gen_val = data_generators.get_anchor_gt(val_imgs,
class_count,
C,
K.image_dim_ordering(),
mode='val')
input_shape_img = (None, None, 3)
img_input = Input(shape=input_shape_img)
roi_input = Input(shape=(C.num_rois, 4))
shared_layers = resnet.ResNet50(img_input, trainable=True)
#RPN built on the base layers
num_anchors = len(C.anchor_box_scales) * len(C.anchor_box_ratios)
rpn = resnet.rpn(shared_layers, num_anchors)
classifier = resnet.classifier(shared_layers,
roi_input,
C.num_rois,
nb_classes=len(class_count),
trainable=True)
model_rpn = Model(img_input, rpn[:2])
model_classifier = Model([img_input, roi_input], classifier)
model_all = Model([img_input, roi_input], rpn[:2] + classifier)
try:
print('loading weights')
model_rpn.load_weights(C.base_net_weight)
model_classifier.load_weights(C.base_net_weight)
except Exception as e:
print(e)
optimizer = Adam(lr=1e-3)
optimizer_classifier = Adam(lr=1e-3)
model_rpn.compile(optimizer=optimizer,
loss=[
model.losses.rpn_loss_cls(num_anchors),
model.losses.rpn_loss_regr(num_anchors)
])
model_classifier.compile(
optimizer=optimizer_classifier,
loss=[
model.losses.class_loss_cls,
model.losses.class_loss_regr(len(class_count) - 1)
],
metrics={'dense_class_{}'.format(len(class_count)): 'accuracy'})
model_all.compile(optimizer='sgd', loss='mae')
epoch_length = 1000
iter_num = 0
losses = np.zeros((epoch_length, 5))
rpn_accuracy_rpn_monitor = []
rpn_accuracy_for_epoch = []
start_time = time.time()
best_loss = np.Inf
class_mapping_inv = inv_map
print('start training')
for epoch_num in range(num_epoches):
progbar = generic_utils.Progbar(epoch_length)
print('Epoch {}/{}'.format(epoch_num + 1, num_epoches))
while True:
try:
if len(rpn_accuracy_rpn_monitor) == epoch_length and C.verbose:
mean_overlapping_bboxes = float(
sum(rpn_accuracy_rpn_monitor)) / len(
rpn_accuracy_rpn_monitor)
rpn_accuracy_rpn_monitor = []
print(
'Average number of overlapping bounding boxes from RPN = {} for {} previous iterations'
.format(mean_overlapping_bboxes, epoch_length))
if mean_overlapping_bboxes == 0:
print(
'RPN is not producing bounding boxes that overlap the ground truth boxes. Check RPN settings or keep training.'
)
X, Y, img_data = data_gen_train.next()
loss_rpn = model_rpn.train_on_batch(X, Y)
P_rpn = model_rpn.predict_on_batch(X)
R = model.roi_helpers.rpn_to_roi(P_rpn[0],
P_rpn[1],
C,
K.image_dim_ordering(),
use_regr=True,
overlap_thresh=0.7,
max_boxes=300)
# note: calc_iou converts from (x1,y1,x2,y2) to (x,y,w,h) format
X2, Y1, Y2 = model.roi_helpers.calc_iou(
R, img_data, C, class_mapping)
if X2 is None:
rpn_accuracy_rpn_monitor.append(0)
rpn_accuracy_for_epoch.append(0)
continue
neg_samples = np.where(Y1[0, :, -1] == 1)
pos_samples = np.where(Y1[0, :, -1] == 0)
if len(neg_samples) > 0:
neg_samples = neg_samples[0]
else:
neg_samples = []
if len(pos_samples) > 0:
pos_samples = pos_samples[0]
else:
pos_samples = []
rpn_accuracy_rpn_monitor.append(len(pos_samples))
rpn_accuracy_for_epoch.append((len(pos_samples)))
if C.num_rois > 1:
if len(pos_samples) < C.num_rois / 2:
selected_pos_samples = pos_samples.tolist()
else:
selected_pos_samples = np.random.choice(
pos_samples, C.num_rois / 2,
replace=False).tolist()
try:
selected_neg_samples = np.random.choice(
neg_samples,
C.num_rois - len(selected_pos_samples),
replace=False).tolist()
except:
selected_neg_samples = np.random.choice(
neg_samples,
C.num_rois - len(selected_pos_samples),
replace=True).tolist()
sel_samples = selected_pos_samples + selected_neg_samples
else:
# in the extreme case where num_rois = 1, we pick a random pos or neg sample
selected_pos_samples = pos_samples.tolist()
selected_neg_samples = neg_samples.tolist()
if np.random.randint(0, 2):
sel_samples = random.choice(neg_samples)
else:
sel_samples = random.choice(pos_samples)
loss_class = model_classifier.train_on_batch(
[X, X2[:, sel_samples, :]],
[Y1[:, sel_samples, :], Y2[:, sel_samples, :]])
losses[iter_num, 0] = loss_rpn[1]
losses[iter_num, 1] = loss_rpn[2]
losses[iter_num, 2] = loss_class[1]
losses[iter_num, 3] = loss_class[2]
losses[iter_num, 4] = loss_class[3]
iter_num += 1
progbar.update(
iter_num,
[('rpn_cls', np.mean(losses[:iter_num, 0])),
('rpn_regr', np.mean(losses[:iter_num, 1])),
('detector_cls', np.mean(losses[:iter_num, 2])),
('detector_regr', np.mean(losses[:iter_num, 0]))])
if iter_num == epoch_length:
loss_rpn_cls = np.mean(losses[:, 0])
loss_rpn_regr = np.mean(losses[:, 1])
loss_class_cls = np.mean(losses[:, 2])
loss_class_regr = np.mean(losses[:, 3])
class_acc = np.mean(losses[:, 4])
mean_overlapping_bboxes = float(sum(
rpn_accuracy_for_epoch)) / len(rpn_accuracy_for_epoch)
rpn_accuracy_for_epoch = []
if C.verbose:
print(
'Mean number of bounding boxes from RPN overlapping ground truth boxes: {}'
.format(mean_overlapping_bboxes))
print(
'Classifier accuracy for bounding boxes from RPN: {}'
.format(class_acc))
print('Loss RPN classifier: {}'.format(loss_rpn_cls))
print('Loss RPN regression: {}'.format(loss_rpn_regr))
print('Loss Detector classifier: {}'.format(
loss_class_cls))
print('Loss Detector regression: {}'.format(
loss_class_regr))
print('Elapsed time: {}'.format(time.time() -
start_time))
curr_loss = loss_rpn_cls + loss_rpn_regr + loss_class_cls + loss_class_regr
iter_num = 0
start_time = time.time()
if curr_loss < best_loss:
if C.verbose:
print(
'Total loss decreased from {} to {}, saving weights'
.format(best_loss, curr_loss))
best_loss = curr_loss
model_all.save_weights(C.model_path)
break
except Exception as e:
print('Exception: {}'.format(e))
continue
print('Training complete, exiting.')