code
stringlengths
66
870k
docstring
stringlengths
19
26.7k
func_name
stringlengths
1
138
language
stringclasses
1 value
repo
stringlengths
7
68
path
stringlengths
5
324
url
stringlengths
46
389
license
stringclasses
7 values
def compute_metrics(self, results: list) -> dict: """Compute the metrics from processed results. Args: results (list): The processed results of each batch. Returns: dict: The computed metrics. The keys are the names of the metrics, and the values are correspo...
Compute the metrics from processed results. Args: results (list): The processed results of each batch. Returns: dict: The computed metrics. The keys are the names of the metrics, and the values are corresponding results.
compute_metrics
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/ava_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/ava_metric.py
Apache-2.0
def process(self, data_batch, data_samples): """Process one batch of data samples. The processed results should be stored in ``self.results``, which will be used to computed the metrics when all batches have been processed. Args: data_batch: A batch of data from the dataloa...
Process one batch of data samples. The processed results should be stored in ``self.results``, which will be used to computed the metrics when all batches have been processed. Args: data_batch: A batch of data from the dataloader. data_samples (Sequence[dict]): A batch ...
process
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multimodal_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multimodal_metric.py
Apache-2.0
def compute_metrics(self, results: List): """Compute the metrics from processed results. Args: results (dict): The processed results of each batch. Returns: Dict: The computed metrics. The keys are the names of the metrics, and the values are corresponding r...
Compute the metrics from processed results. Args: results (dict): The processed results of each batch. Returns: Dict: The computed metrics. The keys are the names of the metrics, and the values are corresponding results.
compute_metrics
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multimodal_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multimodal_metric.py
Apache-2.0
def process(self, data_batch, data_samples) -> None: """transfer tensors in predictions to CPU.""" for sample in data_samples: question_id = sample['question_id'] pred_answer = sample['pred_answer'] result = { 'question_id': int(question_id), ...
transfer tensors in predictions to CPU.
process
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multimodal_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multimodal_metric.py
Apache-2.0
def compute_metrics(self, results: List): """Dump the result to json file.""" mmengine.dump(results, self.file_path) logger = MMLogger.get_current_instance() logger.info(f'Results has been saved to {self.file_path}.') return {}
Dump the result to json file.
compute_metrics
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multimodal_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multimodal_metric.py
Apache-2.0
def process(self, data_batch, data_samples): """Process one batch of data samples. The processed results should be stored in ``self.results``, which will be used to computed the metrics when all batches have been processed. Args: data_batch: A batch of data from the dataloa...
Process one batch of data samples. The processed results should be stored in ``self.results``, which will be used to computed the metrics when all batches have been processed. Args: data_batch: A batch of data from the dataloader. data_samples (Sequence[dict]): A batch ...
process
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multimodal_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multimodal_metric.py
Apache-2.0
def compute_metrics(self, results: List): """Compute the metrics from processed results. Args: results (dict): The processed results of each batch. Returns: Dict: The computed metrics. The keys are the names of the metrics, and the values are corresponding r...
Compute the metrics from processed results. Args: results (dict): The processed results of each batch. Returns: Dict: The computed metrics. The keys are the names of the metrics, and the values are corresponding results.
compute_metrics
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multimodal_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multimodal_metric.py
Apache-2.0
def process(self, data_batch: Sequence[dict], data_samples: Sequence[dict]): """Process one batch of data and predictions. The processed results should be stored in ``self.results``, which will be used to computed the metrics when all batches have been processed. Args: ...
Process one batch of data and predictions. The processed results should be stored in ``self.results``, which will be used to computed the metrics when all batches have been processed. Args: data_batch (Sequence[dict]): A batch of data from the dataloader. predictions (S...
process
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multimodal_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multimodal_metric.py
Apache-2.0
def compute_metrics(self, results: List): """Compute the metrics from processed results. Args: results (list): The processed results of each batch. Returns: Dict: The computed metrics. The keys are the names of the metrics, and the values are corresponding r...
Compute the metrics from processed results. Args: results (list): The processed results of each batch. Returns: Dict: The computed metrics. The keys are the names of the metrics, and the values are corresponding results.
compute_metrics
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multimodal_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multimodal_metric.py
Apache-2.0
def calculate(pred: Union[np.ndarray, torch.Tensor], target: Union[np.ndarray, torch.Tensor], topk: Union[int, Sequence[int]], pred_indices: (bool) = False, target_indices: (bool) = False) -> float: """Calculate the average recall. ...
Calculate the average recall. Args: pred (torch.Tensor | np.ndarray | Sequence): The prediction results. A :obj:`torch.Tensor` or :obj:`np.ndarray` with shape ``(N, M)`` or a sequence of index/onehot format labels. target (torch.Tensor | n...
calculate
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multimodal_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multimodal_metric.py
Apache-2.0
def _format_pred(label, topk=None, is_indices=False): """format various label to List[indices].""" if is_indices: assert isinstance(label, Sequence), \ '`pred` must be Sequence of indices when' \ f' `pred_indices` set to True, but get {type(label)}' for i, sample...
format various label to List[indices].
_format_pred
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multimodal_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multimodal_metric.py
Apache-2.0
def _format_target(label, is_indices=False): """format various label to List[indices].""" if is_indices: assert isinstance(label, Sequence), \ '`target` must be Sequence of indices when' \ f' `target_indices` set to True, but get {type(label)}' for i, sample_gt i...
format various label to List[indices].
_format_target
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multimodal_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multimodal_metric.py
Apache-2.0
def process(self, data_batch: Sequence[Tuple[Any, dict]], data_samples: Sequence[dict]) -> None: """Process one batch of data samples and predictions. The processed results should be stored in ``self.results``, which will be used to compute the metrics when all batches have been ...
Process one batch of data samples and predictions. The processed results should be stored in ``self.results``, which will be used to compute the metrics when all batches have been processed. Args: data_batch (Sequence[Tuple[Any, dict]]): A batch of data from the data...
process
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multisports_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multisports_metric.py
Apache-2.0
def compute_metrics(self, results: list) -> dict: """Compute the metrics from processed results. Args: results (list): The processed results of each batch. Returns: dict: The computed metrics. The keys are the names of the metrics, and the values are correspo...
Compute the metrics from processed results. Args: results (list): The processed results of each batch. Returns: dict: The computed metrics. The keys are the names of the metrics, and the values are corresponding results.
compute_metrics
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/multisports_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/multisports_metric.py
Apache-2.0
def process(self, data_batch: Optional[Dict], data_samples: Sequence[Dict]) -> None: """Process one batch of data samples and data_samples. The processed results should be stored in ``self.results``, which will be used to compute the metrics when all batches have been processed. ...
Process one batch of data samples and data_samples. The processed results should be stored in ``self.results``, which will be used to compute the metrics when all batches have been processed. Args: data_batch (dict, optional): A batch of data from the dataloader. data_sa...
process
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/retrieval_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/retrieval_metric.py
Apache-2.0
def compute_metrics(self, results: List) -> Dict: """Compute the metrics from processed results. Args: results (list): The processed results of each batch. Returns: dict: The computed metrics. The keys are the names of the metrics, and the values are corresp...
Compute the metrics from processed results. Args: results (list): The processed results of each batch. Returns: dict: The computed metrics. The keys are the names of the metrics, and the values are corresponding results.
compute_metrics
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/retrieval_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/retrieval_metric.py
Apache-2.0
def process(self, data_batch: Sequence[Tuple[Any, dict]], predictions: Sequence[dict]) -> None: """Process one batch of data samples and predictions. The processed results should be stored in ``self.results``, which will be used to compute the metrics when all batches have been p...
Process one batch of data samples and predictions. The processed results should be stored in ``self.results``, which will be used to compute the metrics when all batches have been processed. Args: data_batch (Sequence[Tuple[Any, dict]]): A batch of data from the data...
process
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/video_grounding_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/video_grounding_metric.py
Apache-2.0
def compute_metrics(self, results: list) -> dict: """Compute the metrics from processed results. Args: results (list): The processed results of each batch. Returns: dict: The computed metrics. The keys are the names of the metrics, and the values are correspo...
Compute the metrics from processed results. Args: results (list): The processed results of each batch. Returns: dict: The computed metrics. The keys are the names of the metrics, and the values are corresponding results.
compute_metrics
python
open-mmlab/mmaction2
mmaction/evaluation/metrics/video_grounding_metric.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/evaluation/metrics/video_grounding_metric.py
Apache-2.0
def forward(self, x: torch.Tensor) -> torch.Tensor: """Defines the computation performed at every call.""" N, M, T, V, C = x.size() x = x.permute(0, 1, 3, 4, 2).contiguous() if self.data_bn_type == 'MVC': x = self.data_bn(x.view(N, M * V * C, T)) else: x =...
Defines the computation performed at every call.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/aagcn.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/aagcn.py
Apache-2.0
def _make_stem_layer(self) -> None: """Construct the stem layers consists of a conv+norm+act module and a pooling layer.""" self.conv1 = ConvModule( self.in_channels, 64, kernel_size=7, stride=2, padding=3, bias=False, ...
Construct the stem layers consists of a conv+norm+act module and a pooling layer.
_make_stem_layer
python
open-mmlab/mmaction2
mmaction/models/backbones/c2d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/c2d.py
Apache-2.0
def forward(self, x: torch.Tensor) \ -> Union[torch.Tensor, Tuple[torch.Tensor]]: """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: Union[torch.Tensor or Tuple[torch.Tensor]]: The feature of the ...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: Union[torch.Tensor or Tuple[torch.Tensor]]: The feature of the input samples extracted by the backbone.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/c2d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/c2d.py
Apache-2.0
def init_weights(self): """Initiate the parameters either from existing checkpoint or from scratch.""" if isinstance(self.pretrained, str): logger = MMLogger.get_current_instance() logger.info(f'load model from: {self.pretrained}') load_checkpoint(self, self....
Initiate the parameters either from existing checkpoint or from scratch.
init_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/c3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/c3d.py
Apache-2.0
def forward(self, x): """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. the size of x is (num_batches, 3, 16, 112, 112). Returns: torch.Tensor: The feature of the input samples extracted by the backbo...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. the size of x is (num_batches, 3, 16, 112, 112). Returns: torch.Tensor: The feature of the input samples extracted by the backbone.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/c3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/c3d.py
Apache-2.0
def make_divisible(value, divisor, min_value=None, min_ratio=0.9): """Make divisible function. This function rounds the channel number down to the nearest value that can be divisible by the divisor. Args: value (int): The original channel number. divisor (int): The divisor to fully divi...
Make divisible function. This function rounds the channel number down to the nearest value that can be divisible by the divisor. Args: value (int): The original channel number. divisor (int): The divisor to fully divide the channel number. min_value (int, optional): The minimum valu...
make_divisible
python
open-mmlab/mmaction2
mmaction/models/backbones/mobilenet_v2.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mobilenet_v2.py
Apache-2.0
def forward(self, x): """Defines the computation performed at every call. Args: x (Tensor): The input data. Returns: Tensor: The output of the module. """ def _inner_forward(x): if self.use_res_connect: return x + self.conv(x...
Defines the computation performed at every call. Args: x (Tensor): The input data. Returns: Tensor: The output of the module.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/mobilenet_v2.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mobilenet_v2.py
Apache-2.0
def make_layer(self, out_channels, num_blocks, stride, expand_ratio): """Stack InvertedResidual blocks to build a layer for MobileNetV2. Args: out_channels (int): out_channels of block. num_blocks (int): number of blocks. stride (int): stride of the first block. Defa...
Stack InvertedResidual blocks to build a layer for MobileNetV2. Args: out_channels (int): out_channels of block. num_blocks (int): number of blocks. stride (int): stride of the first block. Defaults to 1 expand_ratio (int): Expand the number of channels of the ...
make_layer
python
open-mmlab/mmaction2
mmaction/models/backbones/mobilenet_v2.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mobilenet_v2.py
Apache-2.0
def forward(self, x): """Defines the computation performed at every call. Args: x (Tensor): The input data. Returns: Tensor or Tuple[Tensor]: The feature of the input samples extracted by the backbone. """ x = self.conv1(x) outs = []...
Defines the computation performed at every call. Args: x (Tensor): The input data. Returns: Tensor or Tuple[Tensor]: The feature of the input samples extracted by the backbone.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/mobilenet_v2.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mobilenet_v2.py
Apache-2.0
def _freeze_stages(self): """Prevent all the parameters from being optimized before ``self.frozen_stages``.""" if self.frozen_stages >= 0: self.conv1.eval() for param in self.conv1.parameters(): param.requires_grad = False for i in range(1, self.fr...
Prevent all the parameters from being optimized before ``self.frozen_stages``.
_freeze_stages
python
open-mmlab/mmaction2
mmaction/models/backbones/mobilenet_v2.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mobilenet_v2.py
Apache-2.0
def train(self, mode=True): """Set the optimization status when training.""" super(MobileNetV2, self).train(mode) self._freeze_stages() if mode and self.norm_eval: for m in self.modules(): if isinstance(m, _BatchNorm): m.eval()
Set the optimization status when training.
train
python
open-mmlab/mmaction2
mmaction/models/backbones/mobilenet_v2.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mobilenet_v2.py
Apache-2.0
def make_temporal_shift(self): """Make temporal shift for some layers.""" for m in self.modules(): if isinstance(m, InvertedResidual) and \ len(m.conv) == 3 and m.use_res_connect: m.conv[0] = TemporalShift( m.conv[0], ...
Make temporal shift for some layers.
make_temporal_shift
python
open-mmlab/mmaction2
mmaction/models/backbones/mobilenet_v2_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mobilenet_v2_tsm.py
Apache-2.0
def init_weights(self): """Initiate the parameters either from existing checkpoint or from scratch.""" if self.pretrained2d: logger = MMLogger.get_current_instance() self.load_original_weights(logger) else: if self.pretrained: self.init...
Initiate the parameters either from existing checkpoint or from scratch.
init_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/mobilenet_v2_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mobilenet_v2_tsm.py
Apache-2.0
def make_temporal_shift(self): """Make temporal shift for some layers. To make reparameterization work, we can only build the shift layer before the 'block', instead of the 'blockres' """ def make_block_temporal(stage, num_segments): """Make temporal shift on some b...
Make temporal shift for some layers. To make reparameterization work, we can only build the shift layer before the 'block', instead of the 'blockres'
make_temporal_shift
python
open-mmlab/mmaction2
mmaction/models/backbones/mobileone_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mobileone_tsm.py
Apache-2.0
def make_block_temporal(stage, num_segments): """Make temporal shift on some blocks. Args: stage (nn.Module): Model layers to be shifted. num_segments (int): Number of frame segments. Returns: nn.Module: The shifted blocks. ...
Make temporal shift on some blocks. Args: stage (nn.Module): Model layers to be shifted. num_segments (int): Number of frame segments. Returns: nn.Module: The shifted blocks.
make_block_temporal
python
open-mmlab/mmaction2
mmaction/models/backbones/mobileone_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mobileone_tsm.py
Apache-2.0
def init_weights(self): """Initiate the parameters either from existing checkpoint or from scratch.""" if self.pretrained2d: logger = MMLogger.get_current_instance() self.load_original_weights(logger) else: super().init_weights()
Initiate the parameters either from existing checkpoint or from scratch.
init_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/mobileone_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mobileone_tsm.py
Apache-2.0
def resize_pos_embed(pos_embed: torch.Tensor, src_shape: Tuple[int], dst_shape: Tuple[int], mode: str = 'trilinear', num_extra_tokens: int = 1) -> torch.Tensor: """Resize pos_embed weights. Args: pos_embed (torch.Tensor...
Resize pos_embed weights. Args: pos_embed (torch.Tensor): Position embedding weights with shape [1, L, C]. src_shape (tuple): The resolution of downsampled origin training image, in format (T, H, W). dst_shape (tuple): The resolution of downsampled new training ...
resize_pos_embed
python
open-mmlab/mmaction2
mmaction/models/backbones/mvit.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mvit.py
Apache-2.0
def resize_decomposed_rel_pos(rel_pos: torch.Tensor, q_size: int, k_size: int) -> torch.Tensor: """Get relative positional embeddings according to the relative positions of query and key sizes. Args: rel_pos (Tensor): relative position embeddings (L, C). q_size...
Get relative positional embeddings according to the relative positions of query and key sizes. Args: rel_pos (Tensor): relative position embeddings (L, C). q_size (int): size of query q. k_size (int): size of key k. Returns: Extracted positional embeddings according to rela...
resize_decomposed_rel_pos
python
open-mmlab/mmaction2
mmaction/models/backbones/mvit.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mvit.py
Apache-2.0
def attention_pool(x: torch.Tensor, pool: nn.Module, in_size: Tuple[int], with_cls_token: bool = False, norm: Optional[nn.Module] = None) -> tuple: """Pooling the feature tokens. Args: x (torch.Tensor): The input tensor, should...
Pooling the feature tokens. Args: x (torch.Tensor): The input tensor, should be with shape ``(B, num_heads, L, C)`` or ``(B, L, C)``. pool (nn.Module): The pooling module. in_size (Tuple[int]): The shape of the input feature map. with_cls_token (bool): Whether concatenat...
attention_pool
python
open-mmlab/mmaction2
mmaction/models/backbones/mvit.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/mvit.py
Apache-2.0
def forward(self, x: torch.Tensor) -> torch.Tensor: """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module. """ identity = x out = self.conv1(x) out = s...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet.py
Apache-2.0
def forward(self, x: torch.Tensor) -> torch.Tensor: """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module. """ def _inner_forward(x): """Forward wrapper fo...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet.py
Apache-2.0
def make_res_layer(block: nn.Module, inplanes: int, planes: int, blocks: int, stride: int = 1, dilation: int = 1, style: str = 'pytorch', conv_cfg: Optional[ConfigType] = None, ...
Build residual layer for ResNet. Args: block: (nn.Module): Residual module to be built. inplanes (int): Number of channels for the input feature in each block. planes (int): Number of channels for the output feature in each block. blocks (int): Number of residual blocks. str...
make_res_layer
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet.py
Apache-2.0
def _make_stem_layer(self) -> None: """Construct the stem layers consists of a conv+norm+act module and a pooling layer.""" self.conv1 = ConvModule( self.in_channels, 64, kernel_size=7, stride=2, padding=3, bias=False, ...
Construct the stem layers consists of a conv+norm+act module and a pooling layer.
_make_stem_layer
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet.py
Apache-2.0
def _load_conv_params(conv: nn.Module, state_dict_tv: OrderedDict, module_name_tv: str, loaded_param_names: List[str]) -> None: """Load the conv parameters of resnet from torchvision. Args: conv (nn.Module): The destination conv module. ...
Load the conv parameters of resnet from torchvision. Args: conv (nn.Module): The destination conv module. state_dict_tv (OrderedDict): The state dict of pretrained torchvision model. module_name_tv (str): The name of corresponding conv module in the ...
_load_conv_params
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet.py
Apache-2.0
def _load_bn_params(bn: nn.Module, state_dict_tv: OrderedDict, module_name_tv: str, loaded_param_names: List[str]) -> None: """Load the bn parameters of resnet from torchvision. Args: bn (nn.Module): The destination bn module. stat...
Load the bn parameters of resnet from torchvision. Args: bn (nn.Module): The destination bn module. state_dict_tv (OrderedDict): The state dict of pretrained torchvision model. module_name_tv (str): The name of corresponding bn module in the t...
_load_bn_params
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet.py
Apache-2.0
def _load_torchvision_checkpoint(self, logger: mmengine.MMLogger = None) -> None: """Initiate the parameters from torchvision pretrained checkpoint.""" state_dict_torchvision = _load_checkpoint( self.pretrained, map_location='cpu') if 'state_dict'...
Initiate the parameters from torchvision pretrained checkpoint.
_load_torchvision_checkpoint
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet.py
Apache-2.0
def init_weights(self) -> None: """Initiate the parameters either from existing checkpoint or from scratch.""" if isinstance(self.pretrained, str): logger = MMLogger.get_current_instance() if self.torchvision_pretrain: # torchvision's self....
Initiate the parameters either from existing checkpoint or from scratch.
init_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet.py
Apache-2.0
def forward(self, x: torch.Tensor) \ -> Union[torch.Tensor, Tuple[torch.Tensor]]: """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: Union[torch.Tensor or Tuple[torch.Tensor]]: The feature of the ...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: Union[torch.Tensor or Tuple[torch.Tensor]]: The feature of the input samples extracted by the backbone.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet.py
Apache-2.0
def _freeze_stages(self) -> None: """Prevent all the parameters from being optimized before ``self.frozen_stages``.""" if self.frozen_stages >= 0: self.conv1.bn.eval() for m in self.conv1.modules(): for param in m.parameters(): param.re...
Prevent all the parameters from being optimized before ``self.frozen_stages``.
_freeze_stages
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet.py
Apache-2.0
def train(self, mode: bool = True) -> None: """Set the optimization status when training.""" super().train(mode) self._freeze_stages() if mode and self.norm_eval: for m in self.modules(): if isinstance(m, _BatchNorm): m.eval() if mo...
Set the optimization status when training.
train
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet.py
Apache-2.0
def _freeze_stages(self): """Prevent all the parameters from being optimized before ``self.frozen_stages``.""" if self.frozen_stages >= 0: self.conv1.eval() for param in self.conv1.parameters(): param.requires_grad = False for i in range(1, self.f...
Prevent all the parameters from being optimized before ``self.frozen_stages``.
_freeze_stages
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet2plus1d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet2plus1d.py
Apache-2.0
def forward(self, x): """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The feature of the input samples extracted by the backbone. """ x = self.conv1(x) x = self.maxpoo...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The feature of the input samples extracted by the backbone.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet2plus1d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet2plus1d.py
Apache-2.0
def forward(self, x: torch.Tensor) -> torch.Tensor: """Defines the computation performed at every call.""" def _inner_forward(x): """Forward wrapper for utilizing checkpoint.""" identity = x out = self.conv1(x) out = self.conv2(out) if self....
Defines the computation performed at every call.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def forward(self, x: torch.Tensor) -> torch.Tensor: """Defines the computation performed at every call.""" def _inner_forward(x): """Forward wrapper for utilizing checkpoint.""" identity = x out = self.conv1(x) out = self.conv2(out) out = sel...
Defines the computation performed at every call.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def make_res_layer(block: nn.Module, inplanes: int, planes: int, blocks: int, spatial_stride: Union[int, Sequence[int]] = 1, temporal_stride: Union[int, Sequence[int]] = 1, dilation:...
Build residual layer for ResNet3D. Args: block (nn.Module): Residual module to be built. inplanes (int): Number of channels for the input feature in each block. planes (int): Number of channels for the output feature in each block. ...
make_res_layer
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def _inflate_conv_params(conv3d: nn.Module, state_dict_2d: OrderedDict, module_name_2d: str, inflated_param_names: List[str]) -> None: """Inflate a conv module from 2d to 3d. Args: conv3d (nn.Module): The destination conv3d module. ...
Inflate a conv module from 2d to 3d. Args: conv3d (nn.Module): The destination conv3d module. state_dict_2d (OrderedDict): The state dict of pretrained 2d model. module_name_2d (str): The name of corresponding conv module in the 2d model. inflated...
_inflate_conv_params
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def _inflate_bn_params(bn3d: nn.Module, state_dict_2d: OrderedDict, module_name_2d: str, inflated_param_names: List[str]) -> None: """Inflate a norm module from 2d to 3d. Args: bn3d (nn.Module): The destination bn3d module. s...
Inflate a norm module from 2d to 3d. Args: bn3d (nn.Module): The destination bn3d module. state_dict_2d (OrderedDict): The state dict of pretrained 2d model. module_name_2d (str): The name of corresponding bn module in the 2d model. inflated_param...
_inflate_bn_params
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def _inflate_weights(self, logger: MMLogger) -> None: """Inflate the resnet2d parameters to resnet3d. The differences between resnet3d and resnet2d mainly lie in an extra axis of conv kernel. To utilize the pretrained parameters in 2d model, the weight of conv2d models should be inflate...
Inflate the resnet2d parameters to resnet3d. The differences between resnet3d and resnet2d mainly lie in an extra axis of conv kernel. To utilize the pretrained parameters in 2d model, the weight of conv2d models should be inflated to fit in the shapes of the 3d counterpart. Ar...
_inflate_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def _make_stem_layer(self) -> None: """Construct the stem layers consists of a conv+norm+act module and a pooling layer.""" self.conv1 = ConvModule( self.in_channels, self.base_channels, kernel_size=self.conv1_kernel, stride=(self.conv1_stride_t, s...
Construct the stem layers consists of a conv+norm+act module and a pooling layer.
_make_stem_layer
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def _freeze_stages(self) -> None: """Prevent all the parameters from being optimized before ``self.frozen_stages``.""" if self.frozen_stages >= 0: self.conv1.eval() for param in self.conv1.parameters(): param.requires_grad = False for i in range(1...
Prevent all the parameters from being optimized before ``self.frozen_stages``.
_freeze_stages
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def _init_weights(self, pretrained: Optional[str] = None) -> None: """Initiate the parameters either from existing checkpoint or from scratch. Args: pretrained (str | None): The path of the pretrained weight. Will override the original `pretrained` if set. The arg is...
Initiate the parameters either from existing checkpoint or from scratch. Args: pretrained (str | None): The path of the pretrained weight. Will override the original `pretrained` if set. The arg is added to be compatible with mmdet. Defaults to None.
_init_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def forward(self, x: torch.Tensor) \ -> Union[torch.Tensor, Tuple[torch.Tensor]]: """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor or tuple[torch.Tensor]: The feature of the input s...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor or tuple[torch.Tensor]: The feature of the input samples extracted by the backbone.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def train(self, mode: bool = True) -> None: """Set the optimization status when training.""" super().train(mode) self._freeze_stages() if mode and self.norm_eval: for m in self.modules(): if isinstance(m, _BatchNorm): m.eval()
Set the optimization status when training.
train
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def _freeze_stages(self) -> None: """Prevent all the parameters from being optimized before ``self.frozen_stages``.""" if self.all_frozen: layer = getattr(self, self.layer_name) layer.eval() for param in layer.parameters(): param.requires_grad ...
Prevent all the parameters from being optimized before ``self.frozen_stages``.
_freeze_stages
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def forward(self, x: torch.Tensor) -> torch.Tensor: """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The feature of the input samples extracted by the residual layer. """ r...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The feature of the input samples extracted by the residual layer.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def train(self, mode: bool = True) -> None: """Set the optimization status when training.""" super().train(mode) self._freeze_stages() if mode and self.norm_eval: for m in self.modules(): if isinstance(m, _BatchNorm): m.eval()
Set the optimization status when training.
train
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d.py
Apache-2.0
def train(self, mode=True): """Set the optimization status when training.""" super(ResNet3d, self).train(mode) self._freeze_stages() if mode and self.norm_eval: for m in self.modules(): if isinstance(m, _BatchNorm): m.eval() ...
Set the optimization status when training.
train
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d_csn.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d_csn.py
Apache-2.0
def forward(self, x: torch.Tensor) -> torch.Tensor: """Defines the computation performed at every call.""" # x should be a 5-d tensor assert len(x.shape) == 5 N, C, T, H, W = x.shape out_shape = (N, self.out_channels, self.stride[0] * T, self.stride[1] * H, s...
Defines the computation performed at every call.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d_slowfast.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d_slowfast.py
Apache-2.0
def inflate_weights(self, logger: MMLogger) -> None: """Inflate the resnet2d parameters to resnet3d pathway. The differences between resnet3d and resnet2d mainly lie in an extra axis of conv kernel. To utilize the pretrained parameters in 2d model, the weight of conv2d models should be ...
Inflate the resnet2d parameters to resnet3d pathway. The differences between resnet3d and resnet2d mainly lie in an extra axis of conv kernel. To utilize the pretrained parameters in 2d model, the weight of conv2d models should be inflated to fit in the shapes of the 3d counterpart. For...
inflate_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d_slowfast.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d_slowfast.py
Apache-2.0
def _inflate_conv_params(self, conv3d: nn.Module, state_dict_2d: OrderedDict, module_name_2d: str, inflated_param_names: List[str]) -> None: """Inflate a conv module from 2d to 3d. The differences of conv modules betweene 2d and 3d in Pathway ...
Inflate a conv module from 2d to 3d. The differences of conv modules betweene 2d and 3d in Pathway mainly lie in the inplanes due to lateral connections. To fit the shapes of the lateral connection counterpart, it will expand parameters by concatting conv2d parameters and extra zero pad...
_inflate_conv_params
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d_slowfast.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d_slowfast.py
Apache-2.0
def _freeze_stages(self) -> None: """Prevent all the parameters from being optimized before `self.frozen_stages`.""" if self.frozen_stages >= 0: self.conv1.eval() for param in self.conv1.parameters(): param.requires_grad = False for i in range(1, ...
Prevent all the parameters from being optimized before `self.frozen_stages`.
_freeze_stages
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d_slowfast.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d_slowfast.py
Apache-2.0
def init_weights(self, pretrained: Optional[str] = None) -> None: """Initiate the parameters either from existing checkpoint or from scratch.""" if pretrained: self.pretrained = pretrained # Override the init_weights of i3d super().init_weights() for module_n...
Initiate the parameters either from existing checkpoint or from scratch.
init_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d_slowfast.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d_slowfast.py
Apache-2.0
def build_pathway(cfg: Dict, *args, **kwargs) -> nn.Module: """Build pathway. Args: cfg (dict): cfg should contain: - type (str): identify backbone type. Returns: nn.Module: Created pathway. """ if not (isinstance(cfg, dict) and 'type' in cfg): raise TypeError('...
Build pathway. Args: cfg (dict): cfg should contain: - type (str): identify backbone type. Returns: nn.Module: Created pathway.
build_pathway
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d_slowfast.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d_slowfast.py
Apache-2.0
def init_weights(self, pretrained: Optional[str] = None) -> None: """Initiate the parameters either from existing checkpoint or from scratch.""" if pretrained: self.pretrained = pretrained if isinstance(self.pretrained, str): logger = MMLogger.get_current_instanc...
Initiate the parameters either from existing checkpoint or from scratch.
init_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d_slowfast.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d_slowfast.py
Apache-2.0
def forward(self, x: torch.Tensor) -> tuple: """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: tuple[torch.Tensor]: The feature of the input samples extracted by the backbone. """ x_slow ...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: tuple[torch.Tensor]: The feature of the input samples extracted by the backbone.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet3d_slowfast.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet3d_slowfast.py
Apache-2.0
def forward(self, x: torch.Tensor) -> torch.Tensor: """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module. """ def _inner_forward(x): identity = x ...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_audio.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_audio.py
Apache-2.0
def make_res_layer(block: nn.Module, inplanes: int, planes: int, blocks: int, stride: int = 1, dilation: int = 1, factorize: int = 1, norm_cfg: Optional[Config...
Build residual layer for ResNetAudio. Args: block (nn.Module): Residual module to be built. inplanes (int): Number of channels for the input feature in each block. planes (int): Number of channels for the output feature in each block. ...
make_res_layer
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_audio.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_audio.py
Apache-2.0
def _make_stem_layer(self) -> None: """Construct the stem layers consists of a ``conv+norm+act`` module and a pooling layer.""" self.conv1 = ConvModule( self.in_channels, self.base_channels, kernel_size=self.conv1_kernel, stride=self.conv1_stride, ...
Construct the stem layers consists of a ``conv+norm+act`` module and a pooling layer.
_make_stem_layer
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_audio.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_audio.py
Apache-2.0
def _freeze_stages(self) -> None: """Prevent all the parameters from being optimized before ``self.frozen_stages``.""" if self.frozen_stages >= 0: self.conv1.bn.eval() for m in [self.conv1.conv, self.conv1.bn]: for param in m.parameters(): ...
Prevent all the parameters from being optimized before ``self.frozen_stages``.
_freeze_stages
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_audio.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_audio.py
Apache-2.0
def init_weights(self) -> None: """Initiate the parameters either from existing checkpoint or from scratch.""" if isinstance(self.pretrained, str): logger = MMLogger.get_current_instance() logger.info(f'load model from: {self.pretrained}') load_checkpoint(sel...
Initiate the parameters either from existing checkpoint or from scratch.
init_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_audio.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_audio.py
Apache-2.0
def forward(self, x: torch.Tensor) -> torch.Tensor: """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The feature of the input samples extracted by the backbone. """ x = sel...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The feature of the input samples extracted by the backbone.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_audio.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_audio.py
Apache-2.0
def train(self, mode: bool = True) -> None: """Set the optimization status when training.""" super().train(mode) self._freeze_stages() if mode and self.norm_eval: for m in self.modules(): if isinstance(m, _BatchNorm): m.eval()
Set the optimization status when training.
train
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_audio.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_audio.py
Apache-2.0
def batch_norm(inputs: torch.Tensor, module: nn.modules.batchnorm, training: Optional[bool] = None) -> torch.Tensor: """Applies Batch Normalization for each channel across a batch of data using params from the given batch normalization module. Args: inputs (Tensor): Th...
Applies Batch Normalization for each channel across a batch of data using params from the given batch normalization module. Args: inputs (Tensor): The input data. module (nn.modules.batchnorm): a batch normalization module. Will use params from this batch normalization module to do ...
batch_norm
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_omni.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_omni.py
Apache-2.0
def forward(self, x: torch.Tensor) -> torch.Tensor: """Defines the computation performed at every call. Accept both 3D (BCTHW for videos) and 2D (BCHW for images) tensors. """ if x.ndim == 4: return self.forward_2d(x) # Forward call for 3D tensors. out = sel...
Defines the computation performed at every call. Accept both 3D (BCTHW for videos) and 2D (BCHW for images) tensors.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_omni.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_omni.py
Apache-2.0
def forward(self, x: torch.Tensor) -> torch.Tensor: """Defines the computation performed at every call. Accept both 3D (BCTHW for videos) and 2D (BCHW for images) tensors. """ if x.ndim == 4: return self.forward_2d(x) # Forward call for 3D tensors. x = self....
Defines the computation performed at every call. Accept both 3D (BCTHW for videos) and 2D (BCHW for images) tensors.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_omni.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_omni.py
Apache-2.0
def linear_sampler(data, offset): """Differentiable Temporal-wise Frame Sampling, which is essentially a linear interpolation process. It gets the feature map which has been split into several groups and shift them by different offsets according to their groups. Then compute the weighted sum along ...
Differentiable Temporal-wise Frame Sampling, which is essentially a linear interpolation process. It gets the feature map which has been split into several groups and shift them by different offsets according to their groups. Then compute the weighted sum along with the temporal dimension. Args: ...
linear_sampler
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tin.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tin.py
Apache-2.0
def forward(self, x): """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module. """ # input shape: [num_batches * num_segments, C, H, W] # output x shape: [num_bat...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tin.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tin.py
Apache-2.0
def init_weights(self): """Initiate the parameters either from existing checkpoint or from scratch.""" # we set the initial bias of the convolution # layer to 0, and the final initial output will be 1.0 self.conv.bias.data[...] = 0
Initiate the parameters either from existing checkpoint or from scratch.
init_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tin.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tin.py
Apache-2.0
def forward(self, x): """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module. """ # calculate weight # [N, C, T] n, _, t = x.shape # [N, groups, ...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tin.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tin.py
Apache-2.0
def init_weights(self): """Initiate the parameters either from existing checkpoint or from scratch.""" # The bias of the last fc layer is initialized to # make the post-sigmoid output start from 1 self.fc2.bias.data[...] = 0.5108
Initiate the parameters either from existing checkpoint or from scratch.
init_weights
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tin.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tin.py
Apache-2.0
def forward(self, x): """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module. """ # calculate offset # [N, C, T] n, _, t = x.shape # [N, 1, T] ...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tin.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tin.py
Apache-2.0
def forward(self, x): """Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module. """ # x: [N, C, H, W], # where N = num_batches x num_segments, C = shift_div * num_...
Defines the computation performed at every call. Args: x (torch.Tensor): The input data. Returns: torch.Tensor: The output of the module.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tin.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tin.py
Apache-2.0
def make_temporal_interlace(self): """Make temporal interlace for some layers.""" num_segment_list = [self.num_segments] * 4 assert num_segment_list[-1] > 0 n_round = 1 if len(list(self.layer3.children())) >= 23: print(f'=> Using n_round {n_round} to insert temporal ...
Make temporal interlace for some layers.
make_temporal_interlace
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tin.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tin.py
Apache-2.0
def make_block_interlace(stage, num_segments, shift_div): """Apply Deformable shift for a ResNet layer module. Args: stage (nn.module): A ResNet layer to be deformed. num_segments (int): Number of frame segments. shift_div (int): Number of divisio...
Apply Deformable shift for a ResNet layer module. Args: stage (nn.module): A ResNet layer to be deformed. num_segments (int): Number of frame segments. shift_div (int): Number of division parts for shift. Returns: nn.Sequential: A...
make_block_interlace
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tin.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tin.py
Apache-2.0
def forward(self, x): """Defines the computation performed at every call.""" x = self.block(x) n, c, h, w = x.size() x = x.view(n // self.num_segments, self.num_segments, c, h, w).transpose(1, 2).contiguous() x = self.non_local_block(x) x = x.transpose...
Defines the computation performed at every call.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tsm.py
Apache-2.0
def shift(x, num_segments, shift_div=3): """Perform temporal shift operation on the feature. Args: x (torch.Tensor): The input feature to be shifted. num_segments (int): Number of frame segments. shift_div (int): Number of divisions for shift. Default: 3. Re...
Perform temporal shift operation on the feature. Args: x (torch.Tensor): The input feature to be shifted. num_segments (int): Number of frame segments. shift_div (int): Number of divisions for shift. Default: 3. Returns: torch.Tensor: The shifted feature...
shift
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tsm.py
Apache-2.0
def make_temporal_shift(self): """Make temporal shift for some layers.""" if self.temporal_pool: num_segment_list = [ self.num_segments, self.num_segments // 2, self.num_segments // 2, self.num_segments // 2 ] else: num_segment_...
Make temporal shift for some layers.
make_temporal_shift
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tsm.py
Apache-2.0
def make_block_temporal(stage, num_segments): """Make temporal shift on some blocks. Args: stage (nn.Module): Model layers to be shifted. num_segments (int): Number of frame segments. Returns: nn.Module: The sh...
Make temporal shift on some blocks. Args: stage (nn.Module): Model layers to be shifted. num_segments (int): Number of frame segments. Returns: nn.Module: The shifted blocks.
make_block_temporal
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tsm.py
Apache-2.0
def make_block_temporal(stage, num_segments): """Make temporal shift on some blocks. Args: stage (nn.Module): Model layers to be shifted. num_segments (int): Number of frame segments. Returns: nn.Module: The sh...
Make temporal shift on some blocks. Args: stage (nn.Module): Model layers to be shifted. num_segments (int): Number of frame segments. Returns: nn.Module: The shifted blocks.
make_block_temporal
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tsm.py
Apache-2.0
def make_temporal_pool(self): """Make temporal pooling between layer1 and layer2, using a 3D max pooling layer.""" class TemporalPool(nn.Module): """Temporal pool module. Wrap layer2 in ResNet50 with a 3D max pooling layer. Args: net (nn.Mod...
Make temporal pooling between layer1 and layer2, using a 3D max pooling layer.
make_temporal_pool
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tsm.py
Apache-2.0
def forward(self, x): """Defines the computation performed at every call.""" # [N, C, H, W] n, c, h, w = x.size() # [N // num_segments, C, num_segments, H, W] x = x.view(n // self.num_segments, self.num_segments, c, h, ...
Defines the computation performed at every call.
forward
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tsm.py
Apache-2.0
def make_non_local(self): """Wrap resnet layer into non local wrapper.""" # This part is for ResNet50 for i in range(self.num_stages): non_local_stage = self.non_local_stages[i] if sum(non_local_stage) == 0: continue layer_name = f'layer{i + 1...
Wrap resnet layer into non local wrapper.
make_non_local
python
open-mmlab/mmaction2
mmaction/models/backbones/resnet_tsm.py
https://github.com/open-mmlab/mmaction2/blob/master/mmaction/models/backbones/resnet_tsm.py
Apache-2.0