Shortcuts

Source code for mmaction.models.heads.i3d_head

# Copyright (c) OpenMMLab. All rights reserved.
from mmengine.model.weight_init import normal_init
from torch import Tensor, nn

from mmaction.registry import MODELS
from mmaction.utils import ConfigType
from .base import BaseHead


[docs]@MODELS.register_module() class I3DHead(BaseHead): """Classification head for I3D. Args: num_classes (int): Number of classes to be classified. in_channels (int): Number of channels in input feature. loss_cls (dict or ConfigDict): Config for building loss. Default: dict(type='CrossEntropyLoss') spatial_type (str): Pooling type in spatial dimension. Default: 'avg'. dropout_ratio (float): Probability of dropout layer. Default: 0.5. init_std (float): Std value for Initiation. Default: 0.01. kwargs (dict, optional): Any keyword argument to be used to initialize the head. """ def __init__(self, num_classes: int, in_channels: int, loss_cls: ConfigType = dict(type='CrossEntropyLoss'), spatial_type: str = 'avg', dropout_ratio: float = 0.5, init_std: float = 0.01, **kwargs) -> None: super().__init__(num_classes, in_channels, loss_cls, **kwargs) self.spatial_type = spatial_type self.dropout_ratio = dropout_ratio self.init_std = init_std if self.dropout_ratio != 0: self.dropout = nn.Dropout(p=self.dropout_ratio) else: self.dropout = None self.fc_cls = nn.Linear(self.in_channels, self.num_classes) if self.spatial_type == 'avg': # use `nn.AdaptiveAvgPool3d` to adaptively match the in_channels. self.avg_pool = nn.AdaptiveAvgPool3d((1, 1, 1)) else: self.avg_pool = None
[docs] def init_weights(self) -> None: """Initiate the parameters from scratch.""" normal_init(self.fc_cls, std=self.init_std)
[docs] def forward(self, x: Tensor, **kwargs) -> Tensor: """Defines the computation performed at every call. Args: x (Tensor): The input data. Returns: Tensor: The classification scores for input samples. """ # [N, in_channels, 4, 7, 7] if self.avg_pool is not None: x = self.avg_pool(x) # [N, in_channels, 1, 1, 1] if self.dropout is not None: x = self.dropout(x) # [N, in_channels, 1, 1, 1] x = x.view(x.shape[0], -1) # [N, in_channels] cls_score = self.fc_cls(x) # [N, num_classes] return cls_score