def __init__(self,
input_size: int = 224,
scale_size: int = 256,
normalize: Optional[list] = None,
pre_transform: bool = True,
stage: str = "search",
**kwargs) -> None:
"""
Args:
input_size:
scale_size:
normalize:
pre_transform:
**kwargs:
"""
if normalize is None:
normalize = {'mean': [0.485, 0.456, 0.406], 'std': [0.229, 0.224, 0.225]}
policy = kwargs["policy"]
assert stage in ["search", "train"]
train_epochs = kwargs["hp_policy_epochs"]
self.auto_aug_transform = AutoAugTransform.create(policy, stage=stage, train_epochs=train_epochs)
#self.auto_aug_transform = PbtAutoAugmentClassiferTransform(conf)
if pre_transform:
self.pre_transform = transforms.Resize(input_size)
self.post_transform = transforms.Compose(
transforms=[transforms.Permute(),
transforms.Normalize(**normalize, channel_first=True)],
channel_first=False)
self.cur_epoch = 0
def _init_loader(hparams: dict, TrainTransform=None) -> tuple:
"""
Args:
hparams:
Returns:
"""
train_data_root = hparams.data_config.train_img_prefix
val_data_root = hparams.data_config.val_img_prefix
train_list = hparams.data_config.train_ann_file
val_list = hparams.data_config.val_ann_file
input_size = hparams.task_config.classifier.input_size
scale_size = hparams.task_config.classifier.scale_size
search_space = hparams.search_space
search_space["task_type"] = hparams.task_config.task_type
epochs = hparams.task_config.classifier.epochs
no_cache_img = hparams.task_config.classifier.get("no_cache_img", False)
normalize = {'mean': [0.485, 0.456, 0.406], 'std': [0.229, 0.224, 0.225]}
if TrainTransform is None:
TrainTransform = PbaAugment(
input_size=input_size,
scale_size=scale_size,
normalize=normalize,
policy=search_space,
hp_policy_epochs=epochs,
)
delimiter = hparams.data_config.delimiter
kwargs = dict(conf=hparams, delimiter=delimiter)
if hparams.task_config.classifier.use_class_map:
class_to_id_dict = _read_classes(label_list=hparams.data_config.label_list)
else:
class_to_id_dict = None
train_data = PicReader(
root_path=train_data_root,
list_file=train_list,
transform=TrainTransform,
class_to_id_dict=class_to_id_dict,
cache_img=not no_cache_img,
**kwargs)
val_data = PicReader(
root_path=val_data_root,
list_file=val_list,
transform=transforms.Compose(
transforms=[
transforms.Resize((224, 224)),
transforms.Permute(),
transforms.Normalize(**normalize, channel_first=True)
],
channel_first=False),
class_to_id_dict=class_to_id_dict,
cache_img=not no_cache_img,
**kwargs)
return train_data, val_data
def transforms(self, images: Union[str, np.ndarray]):
transforms = T.Compose([
T.Resize((256, 256)),
T.CenterCrop(224),
T.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
],
to_rgb=True)
return transforms(images).astype('float32')
def valid():
transforms = T.Compose(
[T.Resize((256, 256)),
T.CenterCrop(224),
T.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])],
to_rgb=True)
peach_test = DemoDataset(transforms, mode='test')
model = hub.Module(name='resnet50_vd_imagenet_ssld', label_list=["R0", "B1", "M2", "S3"])
optimizer = paddle.optimizer.Adam(learning_rate=0.001, parameters=model.parameters())
trainer = Trainer(model, optimizer, checkpoint_dir='img_classification_ckpt', use_gpu=True)
trainer.evaluate(peach_test, 16)
def preprocess(image_path):
''' preprocess input image file to np.ndarray
Args:
image_path(str): Path of input image file
Returns:
ProcessedImage(numpy.ndarray): A numpy.ndarray
variable which shape is (1, 3, 192, 192)
'''
transforms = T.Compose([
T.Resize((192, 192)),
T.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
],
to_rgb=True)
return np.expand_dims(transforms(image_path), axis=0)
def preprocess(image_path):
''' preprocess input image file to np.ndarray
Args:
image_path(str): Path of input image file
Returns:
ProcessedImage(numpy.ndarray): A numpy.ndarray
variable which shape is (1, 3, 224, 224)
'''
transforms = T.Compose([
T.Resize((256, 256)),
T.CenterCrop(224),
T.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
],
to_rgb=True)
return np.expand_dims(transforms(image_path), axis=0)
import paddle
import paddlehub as hub
import paddlehub.vision.transforms as T
from paddlehub.finetune.trainer import Trainer
from paddlehub.datasets import Flowers
if __name__ == '__main__':
transforms = T.Compose(
[T.Resize((256, 256)),
T.CenterCrop(224),
T.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])],
to_rgb=True)
flowers = Flowers(transforms)
flowers_validate = Flowers(transforms, mode='val')
model = hub.Module(
name='resnet50_vd_imagenet_ssld',
label_list=["roses", "tulips", "daisy", "sunflowers", "dandelion"],
load_checkpoint=None)
optimizer = paddle.optimizer.Adam(learning_rate=0.001, parameters=model.parameters())
trainer = Trainer(model, optimizer, checkpoint_dir='img_classification_ckpt', use_gpu=True)
trainer.train(flowers, epochs=100, batch_size=32, eval_dataset=flowers_validate, save_interval=10)
def __init__(self, load_checkpoint: str = None):
super(BodyPoseModel, self).__init__()
self.resize_func = P.ResizeScaling()
self.norm_func = T.Normalize(std=[1, 1, 1])
self.pad_func = P.PadDownRight()
self.remove_pad = P.RemovePadding()
self.get_peak = P.GetPeak()
self.get_connection = P.Connection()
self.get_candidate = P.Candidate()
self.draw_pose = P.DrawPose()
no_relu_layers = ['conv5_5_CPM_L1', 'conv5_5_CPM_L2', 'Mconv7_stage2_L1', \
'Mconv7_stage2_L2', 'Mconv7_stage3_L1', 'Mconv7_stage3_L2', \
'Mconv7_stage4_L1', 'Mconv7_stage4_L2', 'Mconv7_stage5_L1', \
'Mconv7_stage5_L2', 'Mconv7_stage6_L1', 'Mconv7_stage6_L1']
blocks = {}
block0 = OrderedDict([('conv1_1', [3, 64, 3, 1, 1]),
('conv1_2', [64, 64, 3, 1, 1]),
('pool1_stage1', [2, 2, 0]),
('conv2_1', [64, 128, 3, 1, 1]),
('conv2_2', [128, 128, 3, 1, 1]),
('pool2_stage1', [2, 2, 0]),
('conv3_1', [128, 256, 3, 1, 1]),
('conv3_2', [256, 256, 3, 1, 1]),
('conv3_3', [256, 256, 3, 1, 1]),
('conv3_4', [256, 256, 3, 1, 1]),
('pool3_stage1', [2, 2, 0]),
('conv4_1', [256, 512, 3, 1, 1]),
('conv4_2', [512, 512, 3, 1, 1]),
('conv4_3_CPM', [512, 256, 3, 1, 1]),
('conv4_4_CPM', [256, 128, 3, 1, 1])])
block1_1 = OrderedDict([('conv5_1_CPM_L1', [128, 128, 3, 1, 1]),
('conv5_2_CPM_L1', [128, 128, 3, 1, 1]),
('conv5_3_CPM_L1', [128, 128, 3, 1, 1]),
('conv5_4_CPM_L1', [128, 512, 1, 1, 0]),
('conv5_5_CPM_L1', [512, 38, 1, 1, 0])])
block1_2 = OrderedDict([('conv5_1_CPM_L2', [128, 128, 3, 1, 1]),
('conv5_2_CPM_L2', [128, 128, 3, 1, 1]),
('conv5_3_CPM_L2', [128, 128, 3, 1, 1]),
('conv5_4_CPM_L2', [128, 512, 1, 1, 0]),
('conv5_5_CPM_L2', [512, 19, 1, 1, 0])])
blocks['block1_1'] = block1_1
blocks['block1_2'] = block1_2
self.model0 = self.make_layers(block0, no_relu_layers)
for i in range(2, 7):
blocks['block%d_1' % i] = OrderedDict([
('Mconv1_stage%d_L1' % i, [185, 128, 7, 1, 3]),
('Mconv2_stage%d_L1' % i, [128, 128, 7, 1, 3]),
('Mconv3_stage%d_L1' % i, [128, 128, 7, 1, 3]),
('Mconv4_stage%d_L1' % i, [128, 128, 7, 1, 3]),
('Mconv5_stage%d_L1' % i, [128, 128, 7, 1, 3]),
('Mconv6_stage%d_L1' % i, [128, 128, 1, 1, 0]),
('Mconv7_stage%d_L1' % i, [128, 38, 1, 1, 0])
])
blocks['block%d_2' % i] = OrderedDict([
('Mconv1_stage%d_L2' % i, [185, 128, 7, 1, 3]),
('Mconv2_stage%d_L2' % i, [128, 128, 7, 1, 3]),
('Mconv3_stage%d_L2' % i, [128, 128, 7, 1, 3]),
('Mconv4_stage%d_L2' % i, [128, 128, 7, 1, 3]),
('Mconv5_stage%d_L2' % i, [128, 128, 7, 1, 3]),
('Mconv6_stage%d_L2' % i, [128, 128, 1, 1, 0]),
('Mconv7_stage%d_L2' % i, [128, 19, 1, 1, 0])
])
for k in blocks.keys():
blocks[k] = self.make_layers(blocks[k], no_relu_layers)
self.model1_1 = blocks['block1_1']
self.model2_1 = blocks['block2_1']
self.model3_1 = blocks['block3_1']
self.model4_1 = blocks['block4_1']
self.model5_1 = blocks['block5_1']
self.model6_1 = blocks['block6_1']
self.model1_2 = blocks['block1_2']
self.model2_2 = blocks['block2_2']
self.model3_2 = blocks['block3_2']
self.model4_2 = blocks['block4_2']
self.model5_2 = blocks['block5_2']
self.model6_2 = blocks['block6_2']
if load_checkpoint is not None:
self.model_dict = paddle.load(load_checkpoint)
self.set_dict(self.model_dict)
print("load custom checkpoint success")
else:
checkpoint = os.path.join(self.directory, 'openpose_body.pdparams')
self.model_dict = paddle.load(checkpoint)
self.set_dict(self.model_dict)
print("load pretrained checkpoint success")
resource_config = config.resource_config
algo_config = config.algo_config
input_size = task_config.classifier.input_size
scale_size = task_config.classifier.scale_size
normalize = {'mean': [0.485, 0.456, 0.406], 'std': [0.229, 0.224, 0.225]}
epochs = task_config.classifier.epochs
policy = args.policy
if policy is None:
print("use normal train transform")
TrainTransform = transforms.Compose(
transforms=[
transforms.Resize((input_size, input_size)),
transforms.Permute(),
transforms.Normalize(**normalize, channel_first=True)
],
channel_first=False)
else:
TrainTransform = PbaAugment(
input_size=input_size,
scale_size=scale_size,
normalize=normalize,
policy=policy,
hp_policy_epochs=epochs,
stage="train")
train_dataset, eval_dataset = _init_loader(config, TrainTransform=TrainTransform)
class_to_id_dict = _read_classes(config.data_config.label_list)
model = hub.Module(
name=config.task_config.classifier.model_name, label_list=class_to_id_dict.keys(), load_checkpoint=None)
def __init__(self, load_checkpoint: str = None):
super(HandPoseModel, self).__init__()
self.norm_func = T.Normalize(std=[1, 1, 1])
self.resize_func = P.ResizeScaling()
self.hand_detect = P.HandDetect()
self.pad_func = P.PadDownRight()
self.remove_pad = P.RemovePadding()
self.draw_pose = P.DrawPose()
self.draw_hand = P.DrawHandPose()
no_relu_layers = ['conv6_2_CPM', 'Mconv7_stage2', 'Mconv7_stage3', \
'Mconv7_stage4', 'Mconv7_stage5', 'Mconv7_stage6']
block1_0 = OrderedDict([('conv1_1', [3, 64, 3, 1, 1]),
('conv1_2', [64, 64, 3, 1, 1]),
('pool1_stage1', [2, 2, 0]),
('conv2_1', [64, 128, 3, 1, 1]),
('conv2_2', [128, 128, 3, 1, 1]),
('pool2_stage1', [2, 2, 0]),
('conv3_1', [128, 256, 3, 1, 1]),
('conv3_2', [256, 256, 3, 1, 1]),
('conv3_3', [256, 256, 3, 1, 1]),
('conv3_4', [256, 256, 3, 1, 1]),
('pool3_stage1', [2, 2, 0]),
('conv4_1', [256, 512, 3, 1, 1]),
('conv4_2', [512, 512, 3, 1, 1]),
('conv4_3', [512, 512, 3, 1, 1]),
('conv4_4', [512, 512, 3, 1, 1]),
('conv5_1', [512, 512, 3, 1, 1]),
('conv5_2', [512, 512, 3, 1, 1]),
('conv5_3_CPM', [512, 128, 3, 1, 1])])
block1_1 = OrderedDict([('conv6_1_CPM', [128, 512, 1, 1, 0]),
('conv6_2_CPM', [512, 22, 1, 1, 0])])
blocks = {}
blocks['block1_0'] = block1_0
blocks['block1_1'] = block1_1
for i in range(2, 7):
blocks['block%d' % i] = OrderedDict([
('Mconv1_stage%d' % i, [150, 128, 7, 1, 3]),
('Mconv2_stage%d' % i, [128, 128, 7, 1, 3]),
('Mconv3_stage%d' % i, [128, 128, 7, 1, 3]),
('Mconv4_stage%d' % i, [128, 128, 7, 1, 3]),
('Mconv5_stage%d' % i, [128, 128, 7, 1, 3]),
('Mconv6_stage%d' % i, [128, 128, 1, 1, 0]),
('Mconv7_stage%d' % i, [128, 22, 1, 1, 0])
])
for k in blocks.keys():
blocks[k] = self.make_layers(blocks[k], no_relu_layers)
self.model1_0 = blocks['block1_0']
self.model1_1 = blocks['block1_1']
self.model2 = blocks['block2']
self.model3 = blocks['block3']
self.model4 = blocks['block4']
self.model5 = blocks['block5']
self.model6 = blocks['block6']
if load_checkpoint is not None:
self.model_dict = paddle.load(load_checkpoint)[0]
self.set_dict(self.model_dict)
print("load custom checkpoint success")
else:
checkpoint = os.path.join(self.directory, 'openpose_hand.pdparams')
self.model_dict = paddle.load(checkpoint)
self.set_dict(self.model_dict)
print("load pretrained checkpoint success")
self.body_model = None
