Cross entropy loss with label smoothing regularizer. Reference: Szegedy et al. Rethinking the Inception Architecture for Computer Vision. CVPR

Triplet loss with hard positive/negative mining. Reference: Hermans et al. In Defense of the Triplet Loss for Person Re-Identification. a

Same, but norm is trainable

r"""Implement of large margin arc distance: : Args: in_features: size of each input sample out_features: size of each

Args: root (string): Base root directory of dataset where [split].pkl and [split].h5 exists split (string): 'train', 'val' or 'te

r"""Set the learning rate of each parameter group using a cosine annealing schedule, where :math:`\eta_{max}` is set to the initial lr, :math:`T_{

Randomly samples N identities each with K instances. Args: data_source (list): contains tuples of (img_path(s), pid, camid). batc

Randomly samples N identities each with K instances. Args: data_source (list): contains tuples of (img_path(s), pid, camid). batc

Cross entropy loss with label smoothing regularizer. Reference: Szegedy et al. Rethinking the Inception Architecture for Computer Vision. CVPR

Base class for bottlenecks that implements `forward()` method.

r"""Set the learning rate of each parameter group using a cosine annealing schedule, where :math:`\eta_{max}` is set to the initial lr, :math:`T_{

r"""Applies a 2D power-average adaptive pooling over an input signal composed of several input planes. The function computed is: :math:`f(X) = pow

Cross entropy loss with label smoothing regularizer. Reference: Szegedy et al. Rethinking the Inception Architecture for Computer Vision. CVPR

Randomly selects a rectangle region in an image and erases its pixels. 'Random Erasing Data Augmentation' by Zhong et al. See https:/

Randomly selects a rectangle region in an image and erases its pixels. 'Random Erasing Data Augmentation' by Zhong et al. See https:/

Randomly selects a rectangle region in an image and erases its pixels. 'Random Erasing Data Augmentation' by Zhong et al. See https:/

Bottleneck for SENet154.

ResNeXt bottleneck type C with a Squeeze-and-Excitation module.

ResNet bottleneck with a Squeeze-and-Excitation module. It follows Caffe implementation and uses `stride=stride` in `conv1` and not in `conv2

Triplet loss with hard positive/negative mining. Reference: Hermans et al. In Defense of the Triplet Loss for Person Re-Identification. a