Computes top-k accuracy, from predicted and true labels. :param scores: scores from the model :param targets: true labels :param k:

get `model` in single-gpu mode or `model.module` in multi-gpu mode.

r"""ResNet-50 model from `"Deep Residual Learning for Image Recognition" <https://arxiv.org/pdf/1512.03385.pdf>`_. Args: pretrained (

1x1 convolution

3x3 convolution with padding

generation multi-leve ROIs upon spatial attention masks with NMS method

r"""ResNet-101 model from `"Deep Residual Learning for Image Recognition" <https://arxiv.org/pdf/1512.03385.pdf>`_. Args: pretrained

3x3 convolution with padding

3x3 convolution with padding

3x3 convolution with padding

Return a sparse matrix used for tensor sketch operation in compact bilinear pooling Args: rand_h: an 1D numpy arr

Only load from one single yaml file. :return: CfgNode

r"""VGG 16-layer model (configuration "D") `"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.

Generate a gaussian blur kernel based on the given radius and std. Args ---------- radius: int Radius of the Gaussian kernel

Wasserstein shaping loss for Bernoulli distribution. Args ---------- assign: torch.cuda.FloatTensor, [batch_size, num_parts, height,

scales: 1D array of anchor sizes in pixels. Example: [32, 64, 128] ratios: 1D array of anchor ratios of width/height. Example: [0.5, 1, 2]

generate default anchor :param anchors_setting: all informations of anchors :param input_shape: shape of input images, e.g. (h, w) :