def models_test(test_status=60, data_type="fabric", only_two_model=None):
# test different score threshold of one-model method without background
one_model = BatchTrain(
cfg_path='../configs/{}/faster_rcnn_r50_fpn_1x_1_model.py'.format(
data_type),
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status)
# test two-model method using one-model method without background as second model
one_model.two_model_test(data_type)
def models_train(test_status=-10, data_type="fabric", const_weights=None):
# train for one-model method without background
BatchTrain(
cfg_path='../configs/{}/faster_rcnn_r50_fpn_1x_1_model.py'.format(
data_type),
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status).common_train()
#
# train first model of two-model method
first_model_train(dataset_name=data_type)
# train constant loss weight of Defect Net method
# train for finding best constant defect finding network loss weight
BatchTrain(
cfg_path='../configs/{}/faster_rcnn_r50_fpn_1x_defectnet_const.py'.
format(data_type),
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status).find_best_constant_loss_weight(
const_weights)
# train variable loss weight of Defect Net method
# train for exponent defect finding network loss weight
BatchTrain(
cfg_path='../configs/{}/faster_rcnn_r50_fpn_1x_defectnet_exp.py'.
format(data_type),
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status).common_train()
#
# train for inverse defect finding network loss weight
BatchTrain(
cfg_path='../configs/{}/faster_rcnn_r50_fpn_1x_defectnet_inv.py'.
format(data_type),
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status).common_train()
#
# train for linear defect finding network loss weight
BatchTrain(
cfg_path='../configs/{}/faster_rcnn_r50_fpn_1x_defectnet_lin.py'.
format(data_type),
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status).common_train()
pass
def models_train(test_status=10, data_type="fabric", const_weights=None):
# train all models
root = '../configs/{}/'.format(data_type)
paths = glob.glob(os.path.join(root, 'baseline_model_*.py'))
paths.sort()
for cfg_path in paths:
m = BatchTrain(cfg_path=cfg_path,
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status)
m.common_train()
# test stacking tricks
paths = glob.glob(os.path.join(root, '*_mst_*k*.py'))
paths.sort()
for cfg_path in paths:
m = BatchTrain(cfg_path=cfg_path,
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status)
params = {
'test_cfg': {
'rcnn':
mmcv.ConfigDict(score_thr=0.05,
nms=dict(type='soft_nms', iou_thr=0.5),
max_per_img=100)
},
'uid': "stacking tricks"
}
def models_test(test_status=60, data_type="fabric", only_two_model=None):
# test different score threshold of one-model method without background
one_model = BatchTrain(
cfg_path='../configs/{}/one_model_cascade_rcnn_r50_fpn_1x.py'.format(
data_type),
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status)
if not only_two_model:
one_model.score_threshold_test()
# test two-model method using one-model method without background as second model
one_model.two_model_test(data_type)
# test different normal images proportion for inverse defect finding network loss weight
if not only_two_model:
BatchTrain(
cfg_path='../configs/{}/defectnet_inverse_cascade_rcnn_r50_fpn_1x.py'
.format(data_type),
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status).normal_proportion_test()
pass
def models_train(test_status=-10, data_type="fabric", const_weights=None):
# train all models
root = '../configs/{}/'.format(data_type)
paths = glob.glob(os.path.join(root, 'defectnet/baseline_model.py'))
paths.sort()
for cfg_path in paths:
m = BatchTrain(cfg_path=cfg_path,
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status)
m.common_train()
# test stacking tricks
paths = glob.glob(os.path.join(root, 'defectnet/*_mst_*_k*.py'))
paths.sort()
for cfg_path in paths:
m = BatchTrain(cfg_path=cfg_path,
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status)
params = {
'test_cfg': {
'rcnn':
mmcv.ConfigDict(score_thr=0.05,
nms=dict(type='soft_nms', iou_thr=0.5),
max_per_img=100)
},
'uid': "stacking tricks"
}
m.common_train(**params)
# test different softnms thresholds
iou_thrs = np.linspace(0.1, 0.9, 9)
for iou_thr in iou_thrs:
softnms_model = BatchTrain(
cfg_path='../configs/{}/baseline_model.py'.format(data_type),
data_mode='test',
train_sleep_time=0,
test_sleep_time=test_status)
params = {
'test_cfg': {
'rcnn':
mmcv.ConfigDict(score_thr=0.05,
nms=dict(type='soft_nms', iou_thr=iou_thr),
max_per_img=100)
},
'uid': iou_thr
}
softnms_model.common_train(**params)