mmselfsup.models.algorithms.simsiam 源代码
# Copyright (c) OpenMMLab. All rights reserved.
from typing import Dict, List, Tuple
import torch
from mmselfsup.registry import MODELS
from mmselfsup.structures import SelfSupDataSample
from .base import BaseModel
[文档]@MODELS.register_module()
class SimSiam(BaseModel):
"""SimSiam.
Implementation of `Exploring Simple Siamese Representation Learning
<https://arxiv.org/abs/2011.10566>`_. The operation of fixing learning rate
of predictor is in `engine/hooks/simsiam_hook.py`.
"""
[文档] def extract_feat(self, inputs: List[torch.Tensor],
**kwarg) -> Tuple[torch.Tensor]:
"""Function to extract features from backbone.
Args:
inputs (List[torch.Tensor]): The input images.
Returns:
Tuple[torch.Tensor]: Backbone outputs.
"""
return self.backbone(inputs[0])
[文档] def loss(self, inputs: List[torch.Tensor],
data_samples: List[SelfSupDataSample],
**kwargs) -> Dict[str, torch.Tensor]:
"""The forward function in training.
Args:
inputs (List[torch.Tensor]): The input images.
data_samples (List[SelfSupDataSample]): All elements required
during the forward function.
Returns:
Dict[str, Tensor]: A dictionary of loss components.
"""
img_v1 = inputs[0]
img_v2 = inputs[1]
z1 = self.neck(self.backbone(img_v1))[0] # NxC
z2 = self.neck(self.backbone(img_v2))[0] # NxC
loss_1 = self.head(z1, z2)
loss_2 = self.head(z2, z1)
losses = dict(loss=0.5 * (loss_1 + loss_2))
return losses