mmpose/models/losses/fea_dis_loss.py

# Copyright (c) OpenMMLab. All rights reserved.
import torch.nn as nn

from mmpose.registry import MODELS


@MODELS.register_module()
class FeaLoss(nn.Module):
    """PyTorch version of feature-based distillation from DWPose Modified from
    the official implementation.

    <https://github.com/IDEA-Research/DWPose>
    Args:
        student_channels(int): Number of channels in the student's feature map.
        teacher_channels(int): Number of channels in the teacher's feature map.
        alpha_fea (float, optional): Weight of dis_loss. Defaults to 0.00007
    """

    def __init__(
        self,
        name,
        use_this,
        student_channels,
        teacher_channels,
        alpha_fea=0.00007,
    ):
        super(FeaLoss, self).__init__()
        self.alpha_fea = alpha_fea

        if teacher_channels != student_channels:
            self.align = nn.Conv2d(
                student_channels,
                teacher_channels,
                kernel_size=1,
                stride=1,
                padding=0)
        else:
            self.align = None

    def forward(self, preds_S, preds_T):
        """Forward function.

        Args:
            preds_S(Tensor): Bs*C*H*W, student's feature map
            preds_T(Tensor): Bs*C*H*W, teacher's feature map
        """

        if self.align is not None:
            outs = self.align(preds_S)
        else:
            outs = preds_S

        loss = self.get_dis_loss(outs, preds_T)

        return loss

    def get_dis_loss(self, preds_S, preds_T):
        loss_mse = nn.MSELoss(reduction='sum')
        N, C, H, W = preds_T.shape

        dis_loss = loss_mse(preds_S, preds_T) / N * self.alpha_fea

        return dis_loss