Retrieve the logger with the specified name or, if name is None, return a logger which is the root logger of the hierarchy. Args:

Get the full path to a checkpoint file. Args: path_to_job (string): the path to the folder of the current job. epoch (int): t

Get the rank of the current process.

Returns the sampler object for the dataset. Args: dataset (Dataset): constructed dataset. split (str):

Retrieves the lr step index for the given epoch. Args: cfg (Config): global config object. cur_epoch (float): the number of

Returns the saved video predictions.

Get the size of the world.

Compute the GPU memory usage for the current device (GB).

Initialize speed set for x and y separately. Initialized speed set is a list of lists [speed_x, speed_y]. First a set o

Interpolated AP - VOCdevkit from VOC 2011.

Generates the label for the MoSI. `separate` label is used for separate prediction on the two axes, i.e., two classifi

Launches the task "func" on one or multiple devices. Args: cfg (Config): global config object. init_method (str): initializa

Load features from path or IO. Args: path (io.BytesIO or string): File path or io.

Entry function for spawning all the function processes.

Convert BN parameters to Sub-BN parameters if model contains Sub-BNs. Args: checkpoint_sd (OrderedDict): source dict of parameters.

Compute the number of parameters. Args: model (model): model to count the number of parameters.

Performs multi-view test on the submission set, where {cfg.TEST.NUM_ENSEMBLE_VIEWS} clips and {cfg.TEST.NUM_SPATIAL_CROPS} crops are sampled

Post processing for videos by multiprocessing. Args: cfg (Config): the global config object. video_list (list): videos name l

Create a RandAugment transform :param config_str: String defining configuration of random augmentation. Consists of multiple sections separat

Standard CutMix bounding-box Generates a random square bbox based on lambda value. This impl includes support for enforcing a border margin a

Min-Max CutMix bounding-box Inspired by Darknet cutmix impl, generates a random rectangular bbox based on min/max percent values applied to e

Reset the metric.

Reset the metric.

Generate different MoSI samples for the data. Args: data (dict): the dictionary that contains a "video" key for the

Compute the temporal intersection over union between a target segment and all the test segments. Parameters ---------- target_segment

Whether the meter logs for ema models or not. Args: model_ema_enabled (bool): indicator of whether ema model

Shuffles the sampler for the dataset. Args: loader (loader): data loader to perform shuffle. cur_epoch (int): n

Soft nms for one video. Args: df: proposals generated by network; alpha: alpha value of Gaussian decaying function; t

Applys static mask with random position and size to the generated pseudo motion sequence Args: frames (Tens

Performs a single optimization step. Args: closure (callable, optional): A closure that reevaluates the model and

Convert the Sub-BN paprameters to normal BN parameters in a state dict. There are two copies of BN layers in a Sub-BN implementation: `bn.bn`

Perform the video training for one epoch. Args: train_loader (loader): video training loader. model (model): the video model

Calculate action classification loss for proposals, but this donot work in epic dataset. Args: preds (dict): predicted action classif

Computes joint MoSI loss. See Ziyuan Huang et al. Self-supervised Motion Learning from Static Images. https://arxiv.org/pdf/2104.0024

Computes only-x MoSI loss. See Ziyuan Huang et al. Self-supervised Motion Learning from Static Images. https://arxiv.org/pdf/2104.002

Computes only-y MoSI loss. See Ziyuan Huang et al. Self-supervised Motion Learning from Static Images. https://arxiv.org/pdf/2104.002

Binary classification confidence maps. Args: preds (dict): predicted classification confidence maps. logits (Tensor): Only fo

Regression confidence maps. Args: preds (dict): predicted regression confidence maps. logits (Tensor): Only for placeholders,

Calculate start and end loss. Args: preds (dict): predicted start and end sequences. logits (Tensor): Only for placeholders,

Args: clip (torch.tensor): Size is (C, T, H, W) Return: clip (torch.tensor): Size is (C, T, H, W)

Args: clip (torch.tensor): Video clip to be cropped. Size is (C, T, H, W) Returns: torch.tensor: randomly

Gets the specified data. Args: index (int): the index of the data in the self._samples list. Returns:

Gets the specified data. Args: index (int): the index of the data in the self._samples list. Returns:

Construct tensors to store the predictions and labels. Expect to get num_clips predictions from each video, and calculate the metrics

Construct tensors to store the predictions and labels. Expect to get num_clips predictions from each video, and calculate the metrics

Args: window_size (int): size of the max length of the deque.

Args: epoch_iters (int): the overall number of iterations of one epoch. cfg (Config): the global config object.

Args: max_iter (int): the max number of iteration of the current epoch. cfg (Config): the global config object.

Args: load (bool): whether or not yaml is needed to be loaded. cfg_dict (dict): dictionary of configs to be updated

When num_folds = 1, MultiFoldDistributedSampler degenerates to DistributedSampler.

Initializes the registry. Args: table_name (str): specifies the name of the registry

We divide each video in epic dataset into multiple sliding windows. Each sliding window is a sample in validation process for

Args: cfg (Config): global config object. split (str): the data split, e.g., "train", "val", "test"

Args: cfg (Config): global config object.

Args: cfg (Config): global config object.

Init Localization dataset. Args: cfg (Configs): global config object. details in utils/config.py split (stri

For initialization of the dataset, the global cfg and the split need to provided. Args: cfg (Config): The global c

Args: c_in (int): number of input channels. ratio (int): reduction ratio for the routing function. kernel

Args: c_in (int): number of input channels. ratio (int): reduction ratio for the routing function. kernel

Args: cfg (Config): global config object. block_idx (list): list of [stage_id, block_id], bot