SEResNet¶
Feature Extraction Link¶
SEResNet¶
-
class
chainercv.links.model.senet.
SEResNet
(n_layer, n_class=None, pretrained_model=None, mean=None, initialW=None, fc_kwargs={})[source]¶ Base class for SE-ResNet architecture.
This architecture is based on ResNet. A squeeze-and-excitation block is applied at the end of each non-identity branch of residual block. Please refer to the original paper for a detailed description of network architecture.
Similar to
chainercv.links.model.resnet.ResNet
, ImageNet pretrained weights are downloaded whenpretrained_model
argument isimagenet
, originally distributed at the Github repository by one of the paper authors.- Parameters
n_layer (int) – The number of layers.
n_class (int) – The number of classes. If
None
, the default values are used. If a supported pretrained model is used, the number of classes used to train the pretrained model is used. Otherwise, the number of classes in ILSVRC 2012 dataset is used.pretrained_model (string) – The destination of the pre-trained chainer model serialized as a
npz
file. If this is one of the strings described above, it automatically loads weights stored under a directory$CHAINER_DATASET_ROOT/pfnet/chainercv/models/
, where$CHAINER_DATASET_ROOT
is set as$HOME/.chainer/dataset
unless you specify another value by modifying the environment variable.mean (numpy.ndarray) – A mean value. If
None
, the default values are used. If a supported pretrained model is used, the mean value used to train the pretrained model is used. Otherwise, the mean value calculated from ILSVRC 2012 dataset is used.initialW (callable) – Initializer for the weights of convolution kernels.
fc_kwargs (dict) – Keyword arguments passed to initialize the
chainer.links.Linear
.
SEResNet50¶
SEResNet101¶
SEResNet152¶
SEResNeXt¶
-
class
chainercv.links.model.senet.
SEResNeXt
(n_layer, n_class=None, pretrained_model=None, mean=None, initialW=None, fc_kwargs={})[source]¶ Base class for SE-ResNeXt architecture.
ResNeXt is a ResNet-based architecture, where grouped convolution is adopted to the second convolution layer of each bottleneck block. In addition, a squeeze-and-excitation block is applied at the end of each non-identity branch of residual block. Please refer to Aggregated Residual Transformations for Deep Neural Networks and Squeeze-and-Excitation Networks for detailed description of network architecture.
Similar to
chainercv.links.model.resnet.ResNet
, ImageNet pretrained weights are downloaded whenpretrained_model
argument isimagenet
, originally distributed at the Github repository by one of the paper authors of SENet.See also
chainercv.links.model.resnet.ResNet
chainercv.links.model.senet.SEResNet
chainercv.links.connection.SEBlock
- Parameters
n_layer (int) – The number of layers.
n_class (int) – The number of classes. If
None
, the default values are used. If a supported pretrained model is used, the number of classes used to train the pretrained model is used. Otherwise, the number of classes in ILSVRC 2012 dataset is used.pretrained_model (string) – The destination of the pre-trained chainer model serialized as a
npz
file. If this is one of the strings described above, it automatically loads weights stored under a directory$CHAINER_DATASET_ROOT/pfnet/chainercv/models/
, where$CHAINER_DATASET_ROOT
is set as$HOME/.chainer/dataset
unless you specify another value by modifying the environment variable.mean (numpy.ndarray) – A mean value. If
None
, the default values are used. If a supported pretrained model is used, the mean value used to train the pretrained model is used. Otherwise, the mean value calculated from ILSVRC 2012 dataset is used.initialW (callable) – Initializer for the weights of convolution kernels.
fc_kwargs (dict) – Keyword arguments passed to initialize the
chainer.links.Linear
.