mmcls.models.utils.batch_augments.mixup 源代码

# Copyright (c) OpenMMLab. All rights reserved.
from typing import Tuple

import numpy as np
import torch

from mmcls.registry import BATCH_AUGMENTS

[文档]@BATCH_AUGMENTS.register_module() class Mixup: r"""Mixup batch augmentation. Mixup is a method to reduces the memorization of corrupt labels and increases the robustness to adversarial examples. It's proposed in `mixup: Beyond Empirical Risk Minimization <>`_ Args: alpha (float): Parameters for Beta distribution to generate the mixing ratio. It should be a positive number. More details are in the note. Note: The :math:`\alpha` (``alpha``) determines a random distribution :math:`Beta(\alpha, \alpha)`. For each batch of data, we sample a mixing ratio (marked as :math:`\lambda`, ``lam``) from the random distribution. """ def __init__(self, alpha: float): assert isinstance(alpha, float) and alpha > 0 self.alpha = alpha
[文档] def mix(self, batch_inputs: torch.Tensor, batch_scores: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]: """Mix the batch inputs and batch one-hot format ground truth. Args: batch_inputs (Tensor): A batch of images tensor in the shape of ``(N, C, H, W)``. batch_scores (Tensor): A batch of one-hot format labels in the shape of ``(N, num_classes)``. Returns: Tuple[Tensor, Tensor): The mixed inputs and labels. """ lam = np.random.beta(self.alpha, self.alpha) batch_size = batch_inputs.size(0) index = torch.randperm(batch_size) mixed_inputs = lam * batch_inputs + (1 - lam) * batch_inputs[index, :] mixed_scores = lam * batch_scores + (1 - lam) * batch_scores[index, :] return mixed_inputs, mixed_scores
[文档] def __call__(self, batch_inputs: torch.Tensor, batch_score: torch.Tensor): """Mix the batch inputs and batch data samples.""" assert batch_score.ndim == 2, \ 'The input `batch_score` should be a one-hot format tensor, '\ 'which shape should be ``(N, num_classes)``.' mixed_inputs, mixed_score = self.mix(batch_inputs, batch_score.float()) return mixed_inputs, mixed_score
Read the Docs v: mmcls-1.x
On Read the Docs
Project Home

Free document hosting provided by Read the Docs.