def R101_512_ff(cfg, pretrained=True):
model = ResNet(Bottleneck, blocks = cfg.MODEL.RESNET.BLOCKS, extras = cfg.MODEL.RESNET.EXTRAS,
se = cfg.MODEL.RESNET.SE, cbam = cfg.MODEL.RESNET.CBAM, fusion = cfg.MODEL.RESNET.FUSION)
if pretrained:
pretrained_dict = load_state_dict_from_url(model_urls['resnet101'])
model_dict = model.state_dict()
pretrained_dict = {k: v for k, v in pretrained_dict.items() if k in model_dict}
model_dict.update(pretrained_dict)
model.load_state_dict(model_dict)
return model
def _shufflenetv2(arch, pretrained, *args, **kwargs):
model = ShuffleNetV2(*args, **kwargs)
if pretrained:
model_url = model_urls[arch]
if model_url is None:
raise NotImplementedError('pretrained {} is not supported as of now'.format(arch))
else:
state_dict = load_state_dict_from_url(model_url)
model.load_state_dict(state_dict, strict=False)
return model
def build_backbone(cfg):
backbone_name = cfg.MODEL.BACKBONE.NAME
print(backbone_name)
if backbone_name == "basic":
model = BasicModel(cfg)
return model
if backbone_name == "vgg":
model = VGG(cfg)
if cfg.MODEL.BACKBONE.PRETRAINED:
state_dict = load_state_dict_from_url(
"https://s3.amazonaws.com/amdegroot-models/vgg16_reducedfc.pth")
model.init_from_pretrain(state_dict)
return model
def SEResnet152_512(cfg, pretrained=True):
model = SEResNet(SEBottleneck,
blocks=cfg.MODEL.RESNET.BLOCKS,
extras=cfg.MODEL.RESNET.EXTRAS)
if pretrained:
pretrained_dict = load_state_dict_from_url(model_urls['resnet152'])
model_dict = model.state_dict()
pretrained_dict = {
k: v
for k, v in pretrained_dict.items() if k in model_dict
}
model_dict.update(pretrained_dict)
model.load_state_dict(model_dict)
return model
def SEResnet34(cfg, pretrained=True):
model = SEResNet(SEBasicBlock,
blocks=cfg.MODEL.RESNET.BLOCKS,
extras=cfg.MODEL.RESNET.EXTRAS,
fusion=cfg.MODEL.RESNET.FUSION)
if pretrained:
pretrained_dict = load_state_dict_from_url(model_urls['resnet34'])
model_dict = model.state_dict()
pretrained_dict = {
k: v
for k, v in pretrained_dict.items() if k in model_dict
}
model_dict.update(pretrained_dict)
model.load_state_dict(model_dict)
return model
def build_backbone(cfg):
backbone_name = cfg.MODEL.BACKBONE.NAME
if backbone_name == "resnet_rdd":
model = ResNetRDD(cfg)
return model
if backbone_name == "resnet_tdt":
model = ResNetTDT(cfg, block=BasicBlock)
return model
if backbone_name == "vgg":
model = VGG(cfg)
if cfg.MODEL.BACKBONE.PRETRAINED:
state_dict = load_state_dict_from_url(
"https://s3.amazonaws.com/amdegroot-models/vgg16_reducedfc.pth"
)
model.init_from_pretrain(state_dict)
return model
def wide_resnet101_2_512(cfg, pretrained=True):
model = ResNet(Bottleneck,
blocks=cfg.MODEL.RESNET.BLOCKS,
extras=cfg.MODEL.RESNET.EXTRAS,
width_per_group=64 * 2)
if pretrained:
pretrained_dict = load_state_dict_from_url(
model_urls['wide_resnet101_2'])
model_dict = model.state_dict()
pretrained_dict = {
k: v
for k, v in pretrained_dict.items() if k in model_dict
}
model_dict.update(pretrained_dict)
model.load_state_dict(model_dict)
return model
def Resnet50_32x4d(cfg, pretrained=True):
model = ResNet(Bottleneck,
blocks=cfg.MODEL.RESNET.BLOCKS,
extras=cfg.MODEL.RESNET.EXTRAS,
groups=32,
width_per_group=4)
if pretrained:
pretrained_dict = load_state_dict_from_url(
model_urls['resnext50_32x4d'])
model_dict = model.state_dict()
pretrained_dict = {
k: v
for k, v in pretrained_dict.items() if k in model_dict
}
model_dict.update(pretrained_dict)
model.load_state_dict(model_dict)
return model
def SEResnet101_32x8d(cfg, pretrained=True):
model = SEResNet(SEBottleneck,
blocks=cfg.MODEL.RESNET.BLOCKS,
extras=cfg.MODEL.RESNET.EXTRAS,
groups=32,
width_per_group=8,
fusion=cfg.MODEL.RESNET.FUSION)
if pretrained:
pretrained_dict = load_state_dict_from_url(
model_urls['resnext101_32x8d'])
model_dict = model.state_dict()
pretrained_dict = {
k: v
for k, v in pretrained_dict.items() if k in model_dict
}
model_dict.update(pretrained_dict)
model.load_state_dict(model_dict)
return model
def build_backbone(cfg):
backbone_name = cfg.MODEL.BACKBONE.NAME
if backbone_name == "Inception":
model = Inception(cfg)
return model
if backbone_name == "basic":
model = BasicModel(cfg)
return model
if backbone_name == "MobileNet":
model = MobileNet(cfg)
return model
if backbone_name == "ResNet50":
model = ResNet50(cfg)
return model
if backbone_name == "ResNet":
model = ResNet(cfg)
return model
if backbone_name == "ResNet50_v2":
model = ResNet50_v2(cfg)
return model
if backbone_name == "ResNext":
model = ResNext(cfg)
return model
if backbone_name == "vgg":
model = VGG(cfg)
if cfg.MODEL.BACKBONE.PRETRAINED:
state_dict = load_state_dict_from_url(
"https://s3.amazonaws.com/amdegroot-models/vgg16_reducedfc.pth")
model.init_from_pretrain(state_dict)
return model
def build_backbone(cfg):
backbone_name = cfg.MODEL.BACKBONE.NAME
print(backbone_name)
if backbone_name == "basic":
model = BasicModel(cfg)
return model
if backbone_name == "vgg":
model = VGG(cfg)
if cfg.MODEL.BACKBONE.PRETRAINED:
state_dict = load_state_dict_from_url(
"https://s3.amazonaws.com/amdegroot-models/vgg16_reducedfc.pth")
model.init_from_pretrain(state_dict)
if backbone_name == "resnet":
depth = cfg.MODEL.BACKBONE.DEPTH
model_urls = {
'resnet18': 'https://download.pytorch.org/models/resnet18-5c106cde.pth',
'resnet34': 'https://download.pytorch.org/models/resnet34-333f7ec4.pth',
'resnet50': 'https://download.pytorch.org/models/resnet50-19c8e357.pth',
'resnet101': 'https://download.pytorch.org/models/resnet101-5d3b4d8f.pth',
'resnet152': 'https://download.pytorch.org/models/resnet152-b121ed2d.pth',}
name_dict = {18: 'resnet18', 34: 'resnet34', 50: 'resnet50', 101: 'resnet101', 152: 'resnet152'}
layers_dict = {18: [2, 2, 2, 2], 34: [3, 4, 6, 3], 50: [3, 4, 6, 3],
101: [3, 4, 23, 3], 152: [3, 8, 36, 3]}
block_dict = {18: BasicBlock, 34: BasicBlock, 50: Bottleneck, 101: Bottleneck, 152: Bottleneck}
model = ResNet(cfg, block_dict[depth], layers_dict[depth])
if cfg.MODEL.BACKBONE.PRETRAINED:
pretrained_dict = model_zoo.load_url(model_urls[name_dict[depth]])
model_dict = model.state_dict()
pretrained_dict = {k: v for k, v in pretrained_dict.items() if k in model_dict}
model_dict.update(pretrained_dict)
model.load_state_dict(model_dict)
return model
if backbone_name == "resnest":
model = ResNest(cfg,BasicBlock)
return model
def vgg(cfg, pretrained=True):
model = VGG(cfg)
if pretrained:
model.init_from_pretrain(load_state_dict_from_url(model_urls['vgg']))
return model
def load_pretrained_weights(model, model_name):
""" Loads pretrained weights, and downloads if loading for the first time. """
state_dict = load_state_dict_from_url(url_map[model_name])
model.load_state_dict(state_dict, strict=False)
print('Loaded pretrained weights for {}'.format(model_name))
def mobilenet_v2(cfg, pretrained=True):
model = MobileNetV2()
if pretrained:
model.load_state_dict(load_state_dict_from_url(model_urls['mobilenet_v2']), strict=False)
return model
def get_vgg(cfg, return_features, norm_func, pretrained, progress, **kwargs):
vgg = VGG(VGG_CONFIG[cfg], return_features, norm_func)
if pretrained:
state_dict = load_state_dict_from_url(kwargs['url'])
vgg.load_state_dict(state_dict, strict=False)
return vgg
